Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberCN102934429 A
Publication typeApplication
Application numberCN 201180023207
PCT numberPCT/JP2011/061288
Publication date13 Feb 2013
Filing date17 May 2011
Priority date18 May 2010
Also published asCN102934429B, CN106331714A, CN106851289A, EP2574055A1, EP2574055A4, US8855434, US9167267, US9253506, US20130028531, US20140355670, US20140355696, US20160044312, WO2011145601A1
Publication number201180023207.5, CN 102934429 A, CN 102934429A, CN 201180023207, CN-A-102934429, CN102934429 A, CN102934429A, CN201180023207, CN201180023207.5, PCT/2011/61288, PCT/JP/11/061288, PCT/JP/11/61288, PCT/JP/2011/061288, PCT/JP/2011/61288, PCT/JP11/061288, PCT/JP11/61288, PCT/JP11061288, PCT/JP1161288, PCT/JP2011/061288, PCT/JP2011/61288, PCT/JP2011061288, PCT/JP201161288
Inventors佐藤数史
Applicant索尼公司
Export CitationBiBTeX, EndNote, RefMan
External Links: SIPO, Espacenet
Image processor and image processing method
CN 102934429 A
Abstract
A deblocking filter (24) performs filtering of decoded image data obtained by decoding image data encoded for each block, so as to remove block distortion. If at least one of two adjacent blocks has an extended block size larger than a given block size where those two blocks are adjacent to each other, a filter setting unit (41) sets the tap length to an extended length to increase the strength of distortion removal, or sets the range of pixels to be filtered to an extended range. When a macroblock having an extended size is used, the degree of smoothing is increased, and pixels including those located away from the block boundary are subjected to filtering. Consequently, even when various block sizes are employed or when blocks of extended sizes are used, images of high image quality can be achieved.
Claims(18)  translated from Chinese
1. 一种图像处理装置,包括: 解码部分,用于对在每个块中的编码的图像数据进行解码; 滤波器,用于对通过所述解码部分解码的解码图像数据施加用于去除块失真的滤波;以及滤波器设置部分,用于根据在块边界处相邻的相邻块的块尺寸设置用于块边界的滤波的抽头长度或者作为滤波对象的滤波对象像素范围。 An image processing apparatus comprising: a decoding section for encoding image data in each block is decoded; filter, decoding section for decoding the decoded image data by applying for removing block distortion filter; and a filter setting section, based on the block boundary for an adjacent block adjacent to the block size setting filter tap length of a filter block boundaries or objects filtered target pixel range.
2.根据权利要求I所述的图像处理装置,其中当将相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将所述抽头长度设置为扩展的长度。 2. The image processing apparatus according to claim I, wherein when the neighboring block when at least one extended block size larger than a predetermined size, the filter tap length setting section setting the length of extension .
3.根据权利要求2所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将滤波器的抽头长度设置得更长。 The image processing apparatus according to claim 2, wherein as the block size increases the adjacent block, the filter setting section to set longer tap length of the filter.
4.根据权利要求I所述的图像处理装置,其中当相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将所述滤波对象像素范围设置为扩展的宽度。 4. The image processing apparatus according to claim I, wherein when at least one of the adjacent block is expanded to a size larger than a predetermined block size, the filter settings section of the filter object pixel range is set to expand the width.
5.根据权利要求4所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将所述滤波对象像素范围设置得更宽。 The image processing apparatus according to claim 4, wherein said adjacent blocks with the block size increases, the filter portion of the filter setting the target pixel set wider range.
6.根据权利要求I所述的图像处理装置,其中当相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将用于滤波的块边界强度数据的值设置为扩展的值。 The image processing apparatus according to claim I, wherein adjacent blocks when at least one extended block size larger than a predetermined size, the filter will be used to set the value of the block boundary strength data filtered as an extension of the value.
7.根据权利要求6所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将用于滤波的块边界强度数据的值设置得更大。 The image processing apparatus according to claim 6, wherein said adjacent blocks with the block size increases, the filter will be used to set the value of the data block boundary filtering strength is set to be larger.
8.根据权利要求I所述的图像处理装置,其中所述滤波器设置部分根据相邻块的相邻边上的块尺寸来设置所述滤波的抽头长度或滤波对象像素范围。 The image processing apparatus according to claim I, wherein said filter setting section setting the tap length of the filter or filtering target pixel block size according to the range of the adjacent edge of the adjacent block.
9.根据权利要求I所述的图像处理装置,其中所述滤波器设置部分根据与相邻块的块尺寸对应的情况分类来设置所述滤波的抽头长度和滤波对象像素范围。 The image processing apparatus according to claim I, wherein said filter setting section based on the case of the adjacent block corresponding to block size classification set tap length of the filter and the filtered target pixel range.
10.根据权利要求9的图像处理装置,其中所述情况分类为所述相邻块都具有预定块尺寸或更小的块尺寸的情况、以及所述相邻块中的至少一个被扩展为大于预定块尺寸的尺寸的情况。 10. The image processing apparatus according to claim 9, wherein the case of the neighboring blocks are classified as blocks having a predetermined size or smaller block size, as well as at least one of said adjacent blocks is greater than the extended when a predetermined size of the block size.
11.根据权利要求10所述的图像处理装置,其中所述滤波器设置部分进行情况分类,以分类为相邻块为16X16像素或更小的情况、两个块中的至少一个大于16X16像素且两个块都为32X32像素或更小的情况、以及两个块中的至少一个大于32X32像素的情况。 The image processing apparatus according to claim 10, wherein the filter setting portion the classification to be classified as an adjacent block 16X16 pixels or smaller, the two blocks at least one larger than 16X16 pixels and Two blocks are 32X32 pixels or less, as well as at least one of the two blocks is larger than 32X32 pixels.
12.根据权利要求I所述的图像处理装置,其中当进行帧内预测或帧间预测时,所述块尺寸是作为处理单位的预测块尺寸。 12. The image processing apparatus according to claim I, wherein when performing inter prediction or intra prediction, the block size as the prediction block size processing unit.
13.根据权利要求I所述的图像处理装置,其中当进行正交变换时,所述块尺寸是作为处理单位的变换尺寸。 The image processing apparatus according to claim I, wherein when the orthogonal transform block size is the size of a conversion processing unit.
14.根据权利要求2所述的图像处理装置,其中所述预定块尺寸是H. 264/AVC标准的宏块尺寸。 14. The image processing apparatus according to claim 2, wherein said predetermined block size H. 264 / AVC standard macro block size.
15.根据权利要求I所述的图像处理装置,其中所述滤波器设置部分根据解码图像数据是用于生成预测图像的图像数据还是用于图像显示的图像数据来设置抽头长度或滤波对象像素范围。 15. The image processing apparatus according to claim I, wherein said filter setting section based on the decoded image data is image data for generating the predicted image or image data for image display to set the target pixel filter tap length or range .
16. —种图像处理方法,包括:解码步骤,对在每个块中的编码的图像数据进行解码; 滤波步骤,对在所述解码步骤中解码的解码图像数据施加用于去除块失真的滤波;以及滤波步骤,根据在块边界处相邻的相邻块的块尺寸设置用于块边界的滤波的抽头长度或者作为滤波对象的滤波对象像素范围。 16. - kinds of image processing method comprising: a decoding step of encoding image data in each block is decoded; filtering step of decoding in the decoding step, decoded image data is applied to the filter for removing block distortion ; and a filtering step, based on the block boundary adjacent block adjacent to the block size setting for the block boundary filtering tap length of an object as a filter or filtering target pixel range.
17. 一种图像处理装置,包括: 滤波器,用于对通过局部解码从正交变换和量化得到的图像数据获得的解码图像数据施加用于去除块失真的滤波; 滤波器设置部分,用于根据在块边界处相邻的相邻块的块尺寸来设置用于块边界的滤波的抽头长度或作为滤波对象的滤波对象像素范围;以及编码部分,用于利用从通过滤波器的滤波得到的解码图像数据在每个图像数据块中进行编码。 17. An image processing apparatus comprising: a filter for the decoded image data is image data obtained by locally decoding the quantized orthogonal transform and the obtained filter for removing block distortion is applied; and a filter setting section for According to the block boundary adjacent block adjacent to the block size is set as a filter tap length or range of the target pixel block boundary filtering for object filtering; and a coding section for use by the filter from the filter to give decoded image data is encoded in each image data block.
18. —种图像处理方法,包括: 滤波步骤,用于对通过局部解码从正交变换和量化得到的图像数据获得的解码图像数据施加用于去除块失真的滤波; 滤波设置步骤,用于根据在块边界处相邻的相邻块的块尺寸来设置用于块边界的滤波的抽头长度或作为滤波对象的滤波对象像素范围;以及编码步骤,用于利用从在所述滤波步骤中的滤波得到的解码图像数据在每个图像数据块中进行编码。 18. - kinds of image processing method, comprising: a filtering step for filtering for removing the block distortion of the decoded image data is image data obtained by locally decoding the quantized orthogonal transform and applying the obtained; filtering setting step, according In the block boundary of adjacent blocks adjacent to the block size setting filter tap length as the target pixel or target range for filtering the block boundary filtering; and an encoding step for use in the filter from the filtering step decoded image data obtained by the image data is encoded in each block.
Description  translated from Chinese

图像处理装置和图像处理方法技术领域[0001] 本发明涉及图像处理装置和图像处理方法,尤其使得可以获得具有优异图像质量的解码图像。 An image processing apparatus and an image processing method TECHNICAL FIELD [0001] The present invention is an image processing apparatus and an image processing method involving, in particular, makes it possible to obtain excellent image quality decoded image. 背景技术[0002] 近年来,在广播站和普通住户中盛行这样的装置,其处理数字的图像信息并高效率地传输或存储信息,或者例如这样的装置,其符合MPEG等的系统,其中通过诸如离散余弦变换的正交变换和运动补偿进行压缩。 [0002] In recent years, the prevalence of such a device in a broadcasting station and ordinary households, which processes the digital image information and efficiently transmit or store information, or for example, an apparatus conforming to MPEG system or the like, wherein by orthogonal transformation such as discrete cosine transform and motion compensation compression. [0003] 特别是,MPEG2 (IS0/IEC 13818-2)被定义为通用图像编码系统,并且目前广泛用于专业使用和用户使用的大范围应用中。 [0003] In particular, MPEG2 (IS0 / IEC 13818-2) it is defined as a generic image coding system, and is now widely used in a wide range of applications for professional use and user use. 在具有例如720X480像素的标准分辨率的隔行扫描图像的情况中,通过分配4到8Mbps的编码量(比特率),使用该MPEG2压缩系统可以获得优异的图像质量。 In the case of an interlaced image having, for example 720X480 pixels in standard resolution, by allocating the amount of 4 to 8Mbps encoding (bit rate), using the MPEG2 compression system can achieve excellent image quality. 在具有1920X1088像素的高分辨率的隔行扫描图像的情况中,通过分配18到22Mbps的编码量(比特率),也可以获得优异的图像质量。 In the case of having a 1920X1088 pixel resolution interlaced image, by allocating the amount of 18 to 22Mbps encoding (bit rate), you can get excellent image quality. [0004] MPEG2旨在用于高图像质量编码以主要适于广播,并且不提供具有较低编码量(比特率)的编码系统,即,高于MPEGl的压缩率。 [0004] MPEG2 is intended for high image quality encoding adapted to the main broadcast, and does not provide a coding system having a lower code amount (bit rate), i.e., higher than MPEGl compression ratio. 随着便携式终端的盛行,预期对该编码系统的需求在未来增大。 With the prevalence of the portable terminal, the expected increase in demand for the coding system in the future. 因此,MPEG4编码系统被标准化。 Thus, MPEG4 encoding system has been standardized. 1998年12月通过了用于图像编码系统的标准作为IS0/IEC 14496-2的国际标准。 December 1998 adopted the standard for image coding system as IS0 / IEC 14496-2 international standard. [0005] 另外,近来,在用于视频会议的图像编码的初始目标下,已经进行对称为H. 26L (ITU-T Q6/16VCEG)的标准的标准化。 [0005] In addition, recently, in the initial target for image coding of a video conference, it has been normalized to a standard called H. 26L (ITU-T Q6 / 16VCEG) a. 已知,H. 26L要求更大量的用于编码和解码的操作, 但是相比于诸如MPEG2、MPEG4等的常规编码系统获得更高的编码效率。 Known, H. 26L requires a larger amount of operations for encoding and decoding, but as compared to MPEG2, MPEG4 and other conventional coding system to achieve higher coding efficiency. 另外,作为MPEG4 的活动的部分,现在在进行基于该H. 26L的用于获得更高编码效率的标准化作为增强压缩视频编码的联合模型。 Further, as part of MPEG4 activities, now carrying the H. 26L based on standardized for higher coding efficiency as a joint model enhanced video compression encoding. 对于标准化安排,在2003年3月建立了名称为H. 264和MPEG-4第10部分(高级视频编码,下文写为“H. 264/AVC”)的国际标准。 For the standardization arrangements in March 2003 to establish a name for H. 264 and MPEG-4 Part 10 (Advanced Video Coding, hereinafter written as "H. 264 / AVC") international standards. [0006] 另外,作为其扩展,在2005年2月完成了对RFExt (保真度范围扩展)的标准化, 其包括商业用途所需的编码工具,诸如RGB、4:2:2和4:4:4以及在MPEG2中定义的8 X 8DCT 和量化矩阵。 [0006] In addition, as part of its expansion, in February 2005 completed the standardization of RFExt (Fidelity Range Extensions), which includes the necessary commercial use coding tools, such as RGB, 4: 2: 2 and 4: 4 : 4 and 8 X 8DCT and quantization matrix defined in MPEG2. 从而,H. 264/AVC系统用作即使在影片中包括胶卷噪声的情况下也能够优异地显示的编码系统,并且用于诸如蓝光(注册商标)的大范围应用中。 Thus, H. 264 / AVC encoding system is used as the system even in the case where the film includes a film excellent in noise can also be displayed, and is used as the Blu-ray (registered trademark), a wide range of applications. [0007] 在这样的编码和解码处理中,以块为单位编码图像数据。 [0007] In such an encoding and decoding process to the encoded image data in units of blocks. 另外,在对编码数据进行解码过程中,如专利文献I所示,例如,通过基于块边界强度和量化参数进行滤波抑制块失真。 Further, in the decoding process of the encoded data, as shown in Patent Document I, for example, by filtering the block distortion suppressed and the block boundary strength based on a quantization parameter. [0008] 另外,近来存在日益增大的用于以更高的压缩比编码的需求,诸如期望压缩约4000X2000像素的图像,或者期望在诸如互联网的具有有限传输能力的环境中分发高分辨率图像。 [0008] In addition, the presence of increasing recently for a higher compression ratio coding needs, such as a desired compression of approximately 4000X2000 pixels, or a desired distribution of high-resolution images, such as environments with limited transmission capacity in the Internet . 从而,在非专利文献I中,提出将宏块的尺寸设置为大于MPEG2或H. 264/AVC的宏块的尺寸,例如32像素X32像素的尺寸。 Thus, in the Non-Patent Document I, the proposed macro block size is set to be greater than MPEG2 or H. 264 / AVC macroblock size, e.g., 32 pixels X32 pixel size. 具体是,在非专利文献I中,对于宏块采用分级结构,从而对于16X 16像素块和更小的块保持与H. 264/AVC中的宏块的兼容性,并将更大的块定义为其超集。 Specifically, in Non-Patent Document I, for the macroblock a hierarchical structure, so that for 16X 16 pixel block and smaller blocks keeping with H. 264 / AVC compatibility in the macro block, and larger block definition its superset. [0009] 现有技术文献[0010] 专利文献[0011] 专利文献I :日本待审专利申请No. 2007-36463[0012] 非专利文献[0013]专利文献 I :“Video Coding Using Extended Block” (StudyGroup 16,Contribution 123,ITU, 2009 年I 月)发明内容[0014] 技术问题[0015] 顺带地,当通过常规去块滤波器去除块失真时,块失真可以尤其是在低比特率下增加,从而,存在未充分去除块失真、且图像质量下降的风险。 [0009] The prior art document [0010] Patent Literature [0011] Patent Document I: Japanese Unexamined Patent Application No. 2007-36463 [0012] Patent Literature [0013] Patent Document I: "Video Coding Using Extended Block" ( StudyGroup 16, Contribution 123, ITU, 2009 年 I month) Summary of the Invention [0014] Technical Problem [0015] Incidentally, when a deblocking filter to remove by conventional block distortion, the block distortion can be, especially at low bit rates increase, Thus, there is a risk not fully removed the block distortion and image degradation. [0016] 因此,本发明技术的目的在于提供一种图像处理装置和图像处理方法,即使在使用各种块尺寸时或者使用扩展尺寸的块时可以提供具有优异图像质量的图像。 [0016] Accordingly, the object of the present technology is to provide an image processing apparatus and an image processing method that can provide an image having excellent image quality, even when using a variety of block sizes or block the use of extended dimensions. [0017] 技术方案[0018] 根据本发明第一方面,提供了一种图像处理装置,包括:解码部分,用于对在每个块中的编码的图像数据进行解码;滤波器,用于对通过所述解码部分解码的解码图像数据施加用于去除块失真的滤波;以及滤波器设置部分,用于根据在块边界处相邻的相邻块的块尺寸设置用于块边界的滤波的抽头长度或者作为滤波对象的滤波对象像素范围。 [0017] Technical Solution [0018] According to a first aspect of the present invention, there is provided an image processing apparatus, comprising: a decoding section for encoding image data in each block is decoded; filter for and a filter tap setting section, in accordance block boundary adjacent block adjacent to the block boundary filtering block size setting used; decoded by the decoding section decoded image data is applied to a filter for removing block distortion length or as a filter object filtering target pixel range. [0019] 在该技术中,提供:滤波器,用于对通过解码在每个块中编码的图像数据获得的解码图像数据施加用于去除块失真的滤波;以及滤波器设置部分,用于设置所述滤波器。 [0019] In this technique, there is provided: filter for the decoded picture data obtained by decoding the encoded image data in each block by the filter for removing block distortion is applied; and a filter setting section for setting said filter. 当在块边界处相邻的相邻块中的至少一个被扩展为大于预定块尺寸时,例如,滤波器设置部分在块尺寸增大时将用于块边界的滤波的抽头长度设置为更长,或者在块尺寸增大时将作为滤波对象的滤波对象像素范围设置为更宽。 When the adjacent block adjacent to the block boundary in at least one extended block size is larger than a predetermined, e.g., a filter setting section when the block size is increased tap length for filtering the block boundary is set to a longer or set as a filter object pixel range filter object when the block size is increased to be wider. 另外,根据相邻块的相邻边的块尺寸设置滤波的抽头长度或滤波对象像素范围。 In addition, according to the block size of the adjacent side of the adjacent block or filter tap length setting target range of pixels filtered. 另外,根据其中相邻块都具有预定块尺寸或更小的尺寸的情况和其中相邻块中的至少一个被扩展为大于预定块尺寸的情况,进行对应于相邻块的块尺寸的情况分类,并且设置滤波的抽头长度和滤波对象像素范围。 Further case, according to which the adjacent blocks are blocks having a predetermined size or smaller size and wherein at least one of the adjacent blocks is expanded to the predetermined block size is larger than the case, carry out the adjacent blocks corresponding to the block size of the classification and set the filter tap length and the filter object pixel range. 进行情况分类,以分类为例如其中相邻块为16X16像素或更小的情况、其中两个块中的至少一个大于16X 16像素且两个块都为32 X 32像素或更小的情况、以及其中两个块中的至少一个大于32 X 32像素的情况。 A case where classification, to classify, for example, 16X16 pixels wherein neighboring blocks is smaller or the case where at least one of the two blocks is greater than 16X 16 pixels and two blocks of 32 X 32 pixels or less, as well as wherein at least one of the two blocks in the case of greater than 32 X 32 pixels. 当进行帧内预测或帧间预测时,块尺寸是作为处理单位的预测块尺寸。 When performing inter prediction or intra prediction, the block size as the prediction block size processing unit. 另外,滤波器设置部分根据解码图像数据是用于生成预测图像的图像数据还是用于图像显示的图像数据设置抽头长度或滤波对象像素范围。 Further, the filter setting section based on the decoded image data is image data for generating the predicted image or image data for image display or the filter tap length setting target pixel range. [0020] 根据本发明第二方面,提供了一种图像处理方法,包括:解码步骤,对在每个块中的编码的图像数据进行解码;滤波步骤,对在所述解码步骤中解码的解码图像数据施加用于去除块失真的滤波;以及滤波步骤,根据在块边界处相邻的相邻块的块尺寸设置用于块边界的滤波的抽头长度或者作为滤波对象的滤波对象像素范围。 [0020] According to a second aspect of the present invention, there is provided an image processing method, comprising: a decoding step of encoding the image data in each block is decoded; filtering step of decoding at the decoding step decodes image data is applied to the filter for removing block distortion; and a filtering step, based on the block boundary adjacent block adjacent to the block size setting for the block boundary filtering tap length of an object as a filter or filtering target pixel range. [0021] 根据本发明第三方面,提供了一种图像处理装置,包括:滤波器,用于对通过局部解码从正交变换和量化得到的图像数据获得的解码图像数据施加用于去除块失真的滤波; 滤波器设置部分,用于根据在块边界处相邻的相邻块的块尺寸来设置用于块边界的滤波的抽头长度或作为滤波对象的滤波对象像素范围;以及编码部分,用于利用从通过滤波器的滤波得到的解码图像数据在每个图像数据块中进行编码。 [0021] According to a third aspect of the present invention, there is provided an image processing apparatus, comprising: a filter for the decoded image data is image data obtained by locally decoding the quantized orthogonal transform obtained and applied for removing block distortion filter; a filter setting section, based on the block boundary for an adjacent block adjacent to the block size used for setting the tap length of the block boundary filtering target pixel range as a filter or filter objects; and encoding section, with on the use of the decoded image data obtained by filtering the filter from the image data encoded in each block. [0022] 根据本发明第四方面,提供了一种图像处理方法,包括:滤波步骤,用于对通过局部解码从正交变换和量化得到的图像数据获得的解码图像数据施加用于去除块失真的滤波;滤波设置步骤,用于根据在块边界处相邻的相邻块的块尺寸来设置用于块边界的滤波的抽头长度或作为滤波对象的滤波对象像素范围;以及编码步骤,用于利用从在所述滤波步骤中的滤波得到的解码图像数据在每个图像数据块中进行编码。 [0022] According to a fourth aspect of the present invention, there is provided an image processing method, comprising: a filtering step for decoding the image data of the image data obtained from the locally decoded orthogonal transform and quantization is applied to obtain for removing block distortion filter; filtering setting step, based on the block boundary for an adjacent block adjacent to the block size setting tap length for the block boundary filtering target pixel range as a filter or filter object; and an encoding step of the use of the decoded image data obtained in the filter filtering step from the encoded image data in each block. [0023] 有利效果[0024] 根据本发明技术,可以获得具有减少的块失真的图像质量优异的图像。 [0023] Advantageous Effects [0024] According to the present invention, techniques can be obtained with reduced image block distortion excellent image quality. 附图说明[0025] 图I示出图像编码装置的配置;[0026] 图2示出在去块滤波器的滤波中使用的像素数据;[0027] 图3示出量化参数QP与阈值α之间的关系;[0028] 图4示出去块滤波器和滤波器设置部分的配置;[0029] 图5示出在图像编码处理中使用的预测块尺寸;[0030] 图6为图像编码处理操作的流程图;[0031] 图7为预测处理的流程图;[0032] 图8为帧内预测处理的流程图;[0033] 图9为帧间预测处理的流程图;[0034] 图10为滤波器设置处理的流程图;[0035] 图11示出图像解码装置的配置;[0036] 图12为图像解码处理操作的流程图;[0037] 图13示出电视装置的示意配置;[0038] 图14示出便携式电话的示意配置;[0039] 图15示出记录和再现装置的示意配置;以及[0040] 图16示出成像装置的示意配置。 BRIEF DESCRIPTION OF DRAWINGS [0025] Figure I shows a configuration of the image coding apparatus; [0026] Figure 2 illustrates the pixel data used in the filtering deblocking filter; [0027] Figure 3 illustrates the quantization parameter QP and the threshold value α Relationship between; [0028] FIG. 4 shows the block out of the filter and the filter configuration setting portion; [0029] FIG. 5 shows a prediction block size used in the image coding process; [0030] FIG. 6 is an image encoding processing operation The flowchart; [0031] FIG. 7 is a flowchart of prediction processing; [0032] FIG. 8 is a flowchart of an intra prediction processing; [0033] FIG. 9 is a flowchart of inter-frame prediction process; [0034] FIG. 10 is filter setting processing flowchart; [0035] FIG. 11 shows a configuration of the image decoding apparatus; [0036] FIG. 12 is a flowchart of image decoding processing operation; [0037] FIG. 13 shows a schematic configuration of a television apparatus; [0038 ] Figure 14 shows a schematic configuration of a portable telephone; [0039] FIG. 15 shows a schematic configuration of a recording and reproducing apparatus; and a schematic configuration [0040] FIG. 16 shows the image forming apparatus. 具体实施方式[0041 ] 下面将描述本发明的具体实施方式。 DETAILED DESCRIPTION [0041] The following will describe specific embodiments of the present invention. 根据本发明的图像处理装置可应用于以预测块尺寸编码图像数据的图像编码装置、用于对以预测块尺寸编码的图像数据进行解码的图像解码装置等等。 The image processing apparatus of the present invention may be applied to the image predictive coding apparatus of the encoded image data block size, the prediction block size used in coding the image data decoding apparatus for decoding an image, and so on. 从而,将按照下面的顺序描述把根据本发明的图像处理装置应用于图像编码装置的情况以及把根据本发明的图像处理装置用于图像解码装置的情况。 Thus, the following sequence describes the case applied to the image coding apparatus as an image processing apparatus according to the present invention and the case of the image decoding apparatus for an image processing apparatus according to the present invention. [0042] I. 图像编码装置的配置[0043] 2. 去块滤波器的滤波[0044] 3. 图像编码装置中的去块滤波器的配置[0045] 4. 图像编码装置的操作[0046] 5. 图像解码装置的配置[0047] 6. 图像解码装置的操作[0048] 7. 应用实例[0049] 〈I.图像编码装置的配置〉[0050] 图I示出一种图像解码装置的配置。 [0042] The image encoding device configuration I. [0043] 2. deblocking filter filter [0044] The image coding apparatus deblocking filter configuration [0045] The image encoding device operation [0046] The image decoding device configuration [0047] The image decoding apparatus operate [0048] 7. Application Examples [0049] <I. image encoding device configuration> [0050] Figure I shows the arrangement of an image decoding apparatus . 图像编码装置10包括模拟/数字转换部分(A/D转换部分)11、画面重排列缓冲器12、减法部分13、正交变换部分14、量化部分15、无损编码部分16、存储缓冲器17、以及速率控制部分18。 Image encoding apparatus 10 includes an analog / digital conversion section (A / D conversion section) 11, picture rearrangement buffer 12, a subtraction portion 13, the orthogonal transform portion 14, quantizing portion 15, the lossless encoding section 16, buffer memory 17, and rate control section 18. 图像编码装置10还包括去量化部分21、反向正交变换部分22、加法部分23、去块滤波器24、帧存储器25、选择器26、帧内预测部分31、运动预测和补偿部分32、以及预测图像和最优模式选择部分33。 Image encoding apparatus 10 further comprises a dequantizing section 21, an inverse orthogonal transformation section 22, addition section 23, a deblocking filter 24, a frame memory 25, the selector 26, the intra prediction section 31, the motion prediction and compensation portion 32, and a prediction image and optimum mode selecting section 33. [0051] A/D转换部分11将模拟图像信号转换为数字图像数据,并将数字图像数据输出到画面重排列缓冲器12。 [0051] A / D conversion section 11 converts the analog image signal into digital image data, and outputs the digital image data to the picture rearrangement buffer 12. [0052] 画面重排列缓冲器12对从A/D转换部分11输出的图像数据的帧进行重排列。 [0052] screen rearrangement buffer 12 converts the frame from A / D section 11 outputs the image data re-arrangement. 画面重排列缓冲器12根据编码处理中涉及的GOP (画面组)结构重排列各个帧,并将重排列后的图像数据输出到减法部分13、帧内预测部分31以及运动预测和补偿部分32。 Picture rearrangement buffer 12 according to the coding of each frame rearrangement process involved GOP (Group of Pictures) structure, and the image data is output to the subtraction section 13 after the rearrangement, the intra prediction section 31 and the motion prediction and compensation portion 32. [0053] 对减法部分13提供从画面重排列缓冲器12输出的图像数据和通过下述预测图像和最优模式选择部分33选择的预测图像数据。 [0053] The subtractor section 13 provides rearrangement buffer 12 outputs the image data from the screen and the selecting section 33 selects the predicted image by the following best mode and predictive image data. 减法部分13计算指示从画面重排列缓冲器12输出的图像数据与从预测图像和最优模式选择部分33提供的预测图像数据之差的预测误差数据,并将预测误差数据输出到正交变换部分14。 Subtracting portion 13 calculates the output buffer 12 indicating the rearranged image data from the screen and select differential prediction error data portion 33 of the predictive image data supplied from the predictive image and the best mode, and predictive error data is output to the orthogonal transform portion 14. [0054] 正交变换部分14对从减法部分13输出的预测误差数据进行正交变换处理,诸如离散余弦变换(DCT)、Karhunen-Loeve变换等等。 [0054] The orthogonal transformation section 14 performs orthogonal transform processing the prediction error data outputted from the subtraction section 13, such as a discrete cosine transform (DCT), Karhunen-Loeve transform and the like. 正交变换部分14将通过进行正交变换处理获得的变换系数数据输出到量化部分15。 Orthogonal transform processing section 14 outputs the obtained transform coefficient data to the quantization section 15 by performing orthogonal transformation. [0055] 对量化部分15提供从正交变换部分14输出的变换系数数据和来自下述的速率控制部分18的速率控制信号。 [0055] The quantization section 15 provided from the orthogonal transform portion 14 and outputs the transform coefficient data rate from the rate control section 18 following the control signal. 量化部分15量化变换系数数据,并将量化数据输出到无损编码部分16和去量化部分21。 Quantizing the quantized transform coefficient data section 15, and outputs quantized data to the lossless encoding section 16 and the dequantization section 21. 另外,量化部分15基于来自速率控制部分18的速率控制信号改变量化参数(量化比例),以改变量化数据的比特率。 Further, based on the quantization part 15 from the speed control section 18 speed control signal changes the quantization parameter (quantization scale), to change the bit rate of the quantized data. [0056] 对无损编码部分16提供从量化部分15输出的量化数据和来自下述的帧内预测部分31、运动预测和补偿部分32以及预测图像和最优模式选择部分33的预测模式信息。 [0056] The lossless encoding section 16 provides the quantized data output from the quantization section 15 and the intra prediction section 31 from below, the motion prediction and compensation portion 32 and the prediction image and the prediction mode optimal mode selection section 33 information. 顺带提及,预测模式信息包括宏块类型,其能够根据帧内预测或帧间预测识别预测块尺寸、预测模式、运动矢量信息、参考画面信息等等。 Incidentally, the prediction mode macroblock type information includes, capable of recognizing the prediction block size, a prediction mode, motion vector information according to the intra prediction or inter prediction, the reference picture information and the like. 无损编码部分16通过例如可变长度编码或算术编码对量化数据进行无损编码处理,从而生成编码流,并将编码流输出到存储缓冲器17。 Lossless encoding section 16 by, for example variable length coding or arithmetic coding on the quantized data lossless encoding processing to generate an encoded stream, the encoded stream output to the memory buffer 17. 另外,无损编码部分16对预测模式信息进行无损编码,并将预测模式信息加到例如编码流的头信息。 In addition, lossless coding mode information prediction section 16 perform lossless encoding and prediction mode information such as header information added to the coded stream. [0057] 存储缓冲器17存储来自无损编码部分16的编码流。 [0057] 17 stores the storage buffer 16 from the lossless encoding section encoding stream. 另外,存储缓冲器17以对应于传输线的传输速度输出存储的编码流。 In addition, the storage buffer 17 corresponding to the transmission speed of the transmission line outputs the stored encoded stream. [0058] 速率控制部分18监视存储缓冲器17的空闲空间,根据空闲空间生成速率控制信号,并将速率控制信号输出到量化部分15。 [0058] The rate control section 18 monitors the free space in the buffer memory 17, a control signal is generated based on the free space velocity, and the rate control signal is output to the quantization section 15. 速率控制部分18例如从存储缓冲器17获得指示空闲空间的信息。 Rate control section 18, for example to obtain information indicating the free space from the storage buffer 17. 当空闲空间减小时,速率控制部分18通过速率控制信号使得量化数据的比特率下降。 When the free space is reduced, the rate control section 18 through the control signal so that the bit rate of the quantized data rate decreases. 当存储缓冲器17具有足够大的空闲空间时,速率控制部分18通过速率控制信号使得量化数据的比特率增高。 When the storage buffer 17 has sufficient free space, the rate control section 18 by the rate control signal so that the bit rate of the quantized data increases. [0059] 去量化部分21对从量化部分15提供的量化数据进行去量化处理。 [0059] dequantizing section 21 quantized data supplied from the quantization section 15 de-quantization. 去量化部分21 将通过进行去量化处理获得的变换系数数据输出到反向正交变换部分22。 Dequantizing section 21 by de-quantization of transform coefficient data obtained is output to the inverse orthogonal transform portion 22. [0060] 反向正交变换部分22将通过对从去量化部分21提供的变换系数数据进行反向正交变换处理获得的数据输出到加法部分23。 [0060] The inverse orthogonal transform portion 22 through the transform coefficient data from the dequantization section 21 provides data obtained by inverse orthogonal transform processing section 23 is output to the adder. [0061] 加法部分23通过将从反向正交变换部分22提供的数据与从预测图像和最优模式选择部分33提供的预测图像数据加到一起生成解码图像数据,并将解码图像数据输出到去块滤波器24和帧存储器25。 [0061] Addition section 23 through the data from the inverse orthogonal transform portion 22 provided with the predictive image data selecting section 33 to provide added together to generate decoded image data from the image and the optimal predictive mode, and outputs the decoded image data to deblocking filter 24 and frame memory 25. [0062] 去块滤波器24进行滤波以减少在图像编码时产生的块失真。 [0062] deblocking filter 24 is filtered to reduce image encoding block distortion generated. 去块滤波器24进行滤波以从自加法部分23提供的解码图像数据中去除块失真,并将滤波后的解码图像数据输出到帧存储器25。 Filtering deblocking filter 24 to remove the block distortion from the decoded picture data from the adding section 23 provided, and the filtered decoded image data is output to the frame memory 25. 另外,去块滤波器24基于从下述滤波器设置部分41提供的参数值设置抽头长度和滤波对象像素范围。 In addition, the deblocking filter 24 based on the setting parameter value to the target pixel filter tap length and scope of section 41 provided from the following filters. [0063] 帧存储器25保存从加法部分23提供的解码图像数据和滤波后的解码图像数据, 所述解码图像数据从去块滤波器24提供。 [0063] The frame memory 25 to save the decoded image data from the addition section 23 provides the decoded image data and the filtering of the decoded image data is provided from a deblocking filter 24. [0064] 选择器26将滤波前的解码图像数据(所述解码图像数据从帧存储器25读取)提供到帧内预测部分31以进行帧内预测。 [0064] selector 26 before filtering the decoded image data (the decoded image data from the frame memory 25 reads) to the intra prediction section 31 for intra prediction. 另外,选择器26将滤波后的解码图像数据(所述解码图像数据从帧存储器25读取)提供到运动预测和补偿部分32以进行帧间预测。 Further, the selector 26 decodes the filtered image data (the decoded image data read from the frame memory 25) is supplied to the motion prediction and compensation portion 32 for inter prediction. [0065] 帧内预测部分31利用从画面重排列缓冲器12输出的编码对象图像的图像数据和滤波前的解码图像数据(该解码图像数据从帧存储器25读取)在所有作为候选的帧内预测模式中进行帧内预测处理。 [0065] the intra prediction section 31 and the decoded image data using the image data before filtering rearrangement buffer 12 output from the picture encoding target picture (the decoded image data read from the frame memory 25) as a candidate in all intra- intra prediction mode prediction process. 另外,帧内预测部分31对于每个帧内预测模式计算成本函数值,并选择其中计算的成本函数值最小的帧内预测模式,即其中获得最高编码效率的帧内预测模式,作为最优帧内预测模式。 Further, the intra prediction section 31 for each intra-prediction mode to calculate the cost function value, and select which to calculate the cost function value the smallest intra-prediction mode, i.e., where the highest coding efficiency of an intra-prediction mode, as the optimal frame intra-prediction mode. 帧内预测部分31将按照最优帧内预测模式生成的预测图像数据、关于最优帧内预测模式的预测模式信息、以及最优帧内预测模式中的成本函数值输出到预测图像和最优模式选择部分33。 Intra prediction section 31 generates the optimal intra prediction mode according to the predicted image data, information about the optimum prediction mode of intra prediction mode, and the cost function value of the optimum intra prediction mode and the optimum prediction image is output to the mode selecting section 33. 另外,为了获得在计算下述的成本函数值中使用的生成编码的量,帧内预测部分31在每个帧内预测模式的帧内预测处理中将关于帧内预测模式的预测模式信息输出到无损编码部分16。 Further, in order to obtain the amount of the calculation of the following cost function value generating encoded, the intra prediction section 31 in the intra prediction processing in each intra-prediction mode on the intra prediction mode information is output to the prediction mode lossless encoding section 16. [0066] 运动预测和补偿部分32在对应于宏块的全部预测块尺寸中进行运动预测和补偿处理。 [0066] The motion prediction and compensation portion 32 performs motion prediction and compensation process in all the prediction block size corresponding to the macro block. 运动预测和补偿部分32利用滤波后的解码图像数据(所述解码图像数据从帧存储器25读取)在从画面重排列缓冲器12读取的编码对象图像中对于各个预测块尺寸的每个图像检测运动矢量。 Motion predicting and decoded image data (the decoded image data read from the frame memory 25) in the screen rearrangement buffer 12 reads from the encoding target image size for each image block prediction compensation portion 32 each use filtered detecting a motion vector. 另外,运动预测和补偿部分32通过基于检测的运动矢量对编码图像施加运动补偿处理生成预测图像。 Further, the motion predicting and compensating section 32 based on the detected motion vector by applying a motion compensation predicted image generated encoded image processing. 另外,运动预测和补偿部分32对于每个预测块尺寸计算成本函数值,并选择其中计算的成本函数值最小的预测块尺寸,即其中获得最高编码效率的预测块尺寸,作为最优帧间预测模式。 Further, the motion prediction and compensation portion 32 of each prediction block size for the calculated cost function values and selecting the value where the calculated cost function minimum prediction block size, i.e. where the highest coding efficiency prediction block size, as the optimal inter prediction mode. 运动预测和补偿部分32将在最优帧间预测模式中生成的预测图像数据、关于最优帧间预测模式的预测模式信息、以及最优帧间预测模式中的成本函数值输出到预测图像和最优模式选择部分33。 Motion prediction and compensation portion 32 generates prediction image data in the optimal inter prediction mode, the prediction mode information about the optimal inter prediction mode, and the cost function value of the optimal inter prediction mode to the prediction image and output Best Mode selecting section 33. 另外,为了获得在计算成本函数值中使用的生成编码的量,运动预测和补偿部分32在每个预测块尺寸的帧间预测处理中将关于帧间预测模式的预测模式信息输出到无损编码部分16。 Further, in order to obtain the amount of the calculated cost function values generated in coding, the motion prediction and compensation portion 32 in the inter-frame prediction process in the prediction block size for each inter prediction mode on the prediction mode information is output to the lossless encoding section 16. 另外,运动预测和补偿部分32按照跳跃宏块和直接模式作为帧间预测模式进行预测。 Further, the motion prediction and compensation portion 32 according to the skip macroblock prediction and direct mode as the inter prediction mode. [0067] 预测图像和最优模式选择部分33以宏块为单位将从帧内预测部分31提供的成本函数值与从运动预测和补偿部分32提供的成本函数值进行比较,并选择较小的成本函数值作为其中获得最佳编码效率的最优模式。 [0067] predicted image selecting section 33 and the best mode macroblock from the intra prediction unit offers cost function value is compared with a portion 31 from the motion prediction and compensation portion 32 provided cost function value, and to select a smaller cost function value as one of the best modes to get the best encoding efficiency. 另外,预测图像和最优模式选择部分33将在最优模式中生成的预测图像数据输出到减法部分13和加法部分23。 Further, the optimal mode prediction image and the prediction image data selecting section 33 will be generated in the optimum mode is output to the subtraction section 13 and addition section 23. 另外,预测图像和最优模式选择部分33将关于最优模式的预测模式信息输出到无损编码部分16和滤波器设置部分41。 Further, the prediction image and optimum mode selecting section 33 with respect to the optimum mode of prediction mode information is output to the lossless encoding section setting section 16 and the filter 41. 另外,预测图像和最优模式选择部分33以片段为单位进行帧内预测或帧间预测。 Further, the prediction image and optimum mode selecting section 33 in units of segments inter prediction or intra prediction. [0068] 滤波器设置部分41根据由关于最优模式的预测模式信息指示的预测块尺寸生成用于设置滤波器的抽头长度和滤波对象像素范围的参数值,并将该参数值输出到去块滤波器24。 [0068] Filter tap length setting portion 41 and filtering the target pixel the parameter value range set according to the filter on the prediction block size is determined by the optimum prediction mode indicated by the mode information is generated for, and the value of the output parameter to deblocking filter 24. [0069] <2.去块滤波器的滤波〉[0070] H264. /AVC的编码系统允许通过在图像压缩信息中包括的画面参数组RBSP的deblocking_f ilter_control_present_f lag 和在片段头中包括的di sab I e_deb locking— filter_idc两个参数指定的去块滤波器的下述三种滤波方法。 [0069] <2. Deblocking filter filter> [0070] H264. / AVC encoding system allows image compression information included in the picture parameter set RBSP of deblocking_f ilter_control_present_f lag and included in the fragment header di sab I The following three filtering methods e_deb locking- filter_idc two parameters specify the deblocking filter. [0071] (a)应用到块边界和太块边界[0072] (b)仅应用到宏块边界[0073] (C)不应用[0074] 对于量化参数QP,当将下面的处理应用到亮度数据时使用QPY,且当将下面的处理应用到色差数据时使用QPC。 [0071] (a) applied to the block boundary and the boundary of the blocks too [0072] (b) applies only to the macro block boundary [0073] (C) does not apply [0074] for the quantization parameter QP, when the following processing is applied to the luminance Use QPY data, and using QPC when the following processing is applied to the color data. 另外,在运动矢量编码、帧内预测和熵编码(CAVLC/CABAC) 中,将属于不同片段的像素值处理为“不可用”。 In addition, the motion vector coding, intra prediction and entropy coding (CAVLC / CABAC), the pixel values belonging to different segments of the processing is "unavailable." 另外,在滤波中,当属于相同的画面时,即使属于不同片段的像素值也被处理为“可用的”。 Further, in the filter, when belonging to the same screen, even if the pixel values belonging to different fragments may also be treated as "available." [0075] 在下面的描述中,假设如图2 (A)所示,在块边界处的滤波前的彼此相邻的块P和Q中的像素数据自边界位置起为PO到p4和qO到q4。 [0075] In the following description, it is assumed in FIG. 2 (A), the pixel data in blocks adjacent to each other P and Q filtering block boundary before the boundary position starting from the PO to p4 and qO to q4. 另外,假设如图2 (B)所示,滤波后的像素数据自边界的位置起为PO'到p4'和qO'到q4'。 Further, it is assumed in FIG. 2 (B), the pixel data of the filtered starting from the position of the boundary of PO 'to p4' and qO 'to q4'. [0076] 在滤波前,如表I所示,对图2中的像素P和像素q定义块边界强度数据Bs (边界强度,Boundary Strength)。 [0076] before filtering, as shown in Table I, Fig. 2 pixel P and pixel q definition block boundary strength data Bs (boundary strength, Boundary Strength). [0077][表 I][0078] [0077] [Table I] [0078]

Figure CN102934429AD00091

[0079] 如表I所示,当像素P和像素q之一属于帧内编码宏块MB、且关注像素位于宏块MB的边界时,对块边界强度数据Bs分配最高滤波强度“4”。 [0079] As shown in Table I, when one pixel P and the pixel q belongs to intra-coded macro block MB, and the pixel of interest is located within the macro block MB boundary of the block boundary strength data Bs is assigned the highest filter strength. "4." [0080] 当像素P和像素q之一属于帧内编码宏块MB、且关注像素不位于宏块MB的边界时,对块边界强度数据Bs分配次于“4”的高滤波强度“3”。 [0080] When one pixel P and the pixel q belongs to intra-coded macro block MB, and the pixel of interest is not located within the macro block MB boundary of the block boundary strength data Bs assignment inferior "4" high filter strength "3" . [0081] 当像素P和像素q都不属于帧内编码宏块MB、且像素之一具有变换系数时,对块边界强度数据Bs分配次于“3”的高滤波强度“2”。 [0081] When the pixel P and the pixel q do not belong to the intra-coded macro block MB, and one pixel has a conversion coefficient for the block boundary strength data Bs allocated inferior "3" high filter strength "2." [0082] 当满足像素P和像素q都不属于帧内编码宏块MB、都不具有变换系数的条件,且满足参考帧不同、参考帧的数目不同、或者运动矢量不同的条件时,对块边界强度数据Bs分配“I”。 [0082] When satisfied pixel P and the pixel q do not belong to the intra-coded macro block MB, do not have the conditions of transform coefficients, and to meet the different reference frames, different number of reference frames, motion vectors or different conditions, on the block distribution boundary strength data Bs "I". [0083] 当像素P和像素q都不属于帧内编码宏块MB、且都不具有变换系数,且参考帧和运动矢量相同时,对块边界强度数据Bs分配“O”。 [0083] When the pixel P and the pixel q do not belong to the intra-coded macro block MB, and they do not have a conversion coefficient, and the reference frame and the motion vectors are the same, to assign the block boundary strength data Bs "O". 另外,“O”表示不进行滤波[0084] 仅当公式(I)中的条件满足时,对图2中的(p2,pl,p0,qO, ql, q2)滤波。 In addition, "O" means no filter [0084] Only when the formula (I) of the conditions are met, to FIG. 2 (p2, pl, p0, qO, ql, q2) filtering. [0085] Bs > 0[0086] p0-q0 < α ;|ρ1_ρθ| < β ; | ql_q0 | < β[0087] ... (I)[0088] 如下根据量化参数QP缺省确定作为用于调节滤波强度(即滤波的容易程度)的参数值的阈值α和β。 [0085] Bs> 0 [0086] p0-q0 <α; | ρ1_ρθ | <β; | ql_q0 | <β [0087] ... (I) [0088] As determined as follows according to the quantization parameter QP is used to adjust the default thresholds α and β filtering strength (i.e., ease of filtering) parameter values. 另外,用户可以通过在图像压缩信息中的片段头中包括的两个参数slice_alpha_c0_offset_div2 和slice_beta_offset_div2 调节强度。 In addition, users can compress the image information in the segment header includes two parameters slice_alpha_c0_offset_div2 and slice_beta_offset_div2 adjust the intensity. 另外,图3 不出量化参数QP与阈值α之间的关系。 In addition, the relationship between the quantization parameter QP 3 no threshold value α between. 当将偏移量加到量化参数QP时,指示量化参数QP与阈值α之间的关系的曲线在箭头方向移动。 When the offset to the quantization parameter QP, indicating the quantization parameter QP curvilinear relationship between a threshold value α is moved in the direction of the arrow. 从而显而易见,调节了滤波强度。 Obviously thereby adjusting the filter strength. [0089] 另外,利用彼此相邻的块P和块Q的各自的量化参数qPp和qPq从公式(2)到(4) 计算indexA和indexB,并且从表2所示的表格获得阈值α和β。 [0089] In addition, the use of their quantization parameter qPp and qPq blocks adjacent to each other and block P and Q are calculated indexA indexB from equation (2) to (4), and the threshold value α and β obtained from the table shown in Table 2 . [0090] qPav = (qPp+qPq+1) >> I ... (2)[0091] indexA=Clip3 (0,51, qPav+FilterOffsetA)…(3)[0092] indexB=Clip3 (0,51, qPav+FilterOffsetB)…(4)[0093][表 2] [0090] qPav = (qPp + qPq + 1) >> I ... (2) [0091] indexA = Clip3 (0,51, qPav + FilterOffsetA) ... (3) [0092] indexB = Clip3 (0,51 , qPav + FilterOffsetB) ... (4) [0093] [Table 2]

Figure CN102934429AD00101

[0097] 首先对其中“Bs〈4”的情况进行描述。 [0097] First, the case where "Bs <4" will be described. [0098] 去块滤波器进行公式(5)到(7)中所示的操作以在滤波后计算像素数据PO'和qO,。 [0098] deblocking filter equation (5) to (7) to the operation shown in calculating the filtered pixel data PO 'and qO ,. [0099] 公式(7)中的Clip3表示裁切处理。 [0099] Equation (7) indicates that the crop processing Clip3. [0100] p0,= Clipl (ρ0+ Δ ) · · · (5)[0101] qO,= Clipl (qO+ Δ ) · · · (6)[0102] Δ = Clip3(-tc, tc((((q0-p O) << 2) + (pl-ql)+4) >> 3))[0103] · · · (7)[0104] 当“chromaEdgeFlag”指示“O”时,去块滤波器基于公式(8)计算公式(7)中的“tc”,否则基于公式(9)计算公式(7)中的“tc”。 [0100] p0, = Clipl (ρ0 + Δ) · · · (5) [0101] qO, = Clipl (qO + Δ) · · · (6) [0102] Δ = Clip3 (-tc, tc ((((q0 -p O) << 2) + (pl-ql) +4) >> 3)) [0103] · · · (7) [0104] When "chromaEdgeFlag" indication "O", the deblocking filter based on the formula (8) formula (7) "tc", or based on the formula (9) formula (7) in the "tc". [0105][0106][0107][0108] [0109]在公式(8)中,当满足括号内的条件时,“ O ?1:0”指示“ I ”,否则指示“O”。 [0105] [0106] [0107] [0108] [0109] In the equation (8), when the condition brackets, "O 1:? 0" indicates "I", or indication "O". tc = tcO+ ((ap < β ) ? 1:0) + (aq < β ) ? I :0) · · · (8) tc = tcO+1 · · · (9)根据Bs和indexA的值,如表3中定义tc的值。 tc = tcO + ((ap <β) 1:? 0) + (aq <β) I:? 0) · · · (8) tc = tcO + 1 · · · (9) the value of Bs and indexA based, such as defined in Table 3 The tc. [表3] [Table 3]

Figure CN102934429AD00111

[0111] 另外,去块滤波器根据公式(10 )和(11)计算公式(8 )中的ap和aq。 [0111] In addition, the deblocking filter according to equation (10) and (11) of formula (8) ap and aq. [0112] ap = |ρ2_ρθ| · · · (10)[0113] aq = | q2_q0 | · · · (11)[0114] 当ChiOmaEdgeFlag为“0”,且ap等于或小于“ β ”时,去块滤波器通过进行公式(12)中所示的运算来计算滤波后的像素数据ρΓ,否则通过公式(13)获得滤波后的像素数据Pi,。 [0112] ap = | ρ2_ρθ | · · · (10) [0113] aq = | q2_q0 | · · · (11) [0114] When ChiOmaEdgeFlag is "0", and ap is equal to or less than "β", the deblocking filter by performing the equation (12) shown in operation to calculate pixel data ρΓ filtered, or by the formula (13) to obtain the pixel data filtered Pi ,. [0115] pi,= pl+Clip3 (_tc0, tcO, (p2+ ((p0+q0+l) > > I)-(p I < < I)) > > I)• · · (12)[0116] pi' = pi · · · (13)[0117] 当chromaEdgeFlag为“0”,且aq等于或小于“ β ”时,去块滤波器通过进行公式(14)中所示的运算来计算滤波后的像素数据ql',否则通过公式(15)获得滤波后的像素数据ql,。 [0115] pi, = pl + Clip3 (_tc0, tcO, (p2 + ((p0 + q0 + l)>> I) - (p I <<I))>> I) • · · (12) [0116] pi '= pi · · · (13) [0117] When chromaEdgeFlag is "0", and aq equal to or less than "β", deblocking operation carried out by the equation (14) shown in filters to calculate the filtered pixel data ql ', or filtered pixel data obtained by the equation (15) after ql ,. [0118] ql,= ql+Clip3 (_tc0, tcO, (q2+ ((p0+q0+l) > > I) - (ql < < I)) > > I) • · · (14)[0119] ql, = ql · · · (15)[0120] 另外,像素数据p2'和像素数据q2'是滤波前的值。 [0118] ql, = ql + Clip3 (_tc0, tcO, (q2 + ((p0 + q0 + l)>> I) - (ql <<I))>> I) • · · (14) [0119] ql , = ql · · · (15) [0120] In addition, the pixel data p2 'and the pixel data q2' is the value before filtering. [0121] ρ2' = ρ2 · · · (16)[0122] q2' = q2 · · · (17)[0123] 下面对其中“Bs=4”的情况进行描述。 [0121] ρ2 '= ρ2 · · · (16) [0122] q2' = q2 · · · (17) [0123] Next, where "Bs = 4" to describe the situation. [0124] 当ChiOmaEdgeFlag指示“O”且满足公式(18)的条件时,去块滤波器根据公式(19) 到(21)计算像素数据pO '、P I'和p2 '。 [0124] When ChiOmaEdgeFlag indication "O" and satisfy the following formula (18) conditions, the deblocking filter according to equation (19) to (21) calculates the pixel data pO ', P I' and p2 '. [0125] ap < β &&Ip0-q0 < ((α >> 2)+2) · · · (18)[0126] pO' = (ρ2+2 · pl+2 · ρΟ+2 · qO+ql+4) >> 3 · · · (19)[0127] ρΓ = (p2+pl+p0+q0+2) >>2 · · · (20)[0128] ρ2' = (2 · ρ3+3 · p2+pl+p0+q0+4) >>3 · · · (21)[0129] 当ChiOmaEdgeFlag指示“O”且不满足公式(18)的条件时,去块滤波器根据公式(22) IlJ (24)计算像素数据pO '、P I' 和p2 '。 [0125] ap <β && Ip0-q0 <((α >> 2) +2) · · · (18) [0126] pO '= (ρ2 + 2 · pl + 2 · ρΟ + 2 · qO + ql + 4 ) >> 3 · · · (19) [0127] ρΓ = (p2 + pl + p0 + q0 + 2) >> 2 · · · (20) [0128] ρ2 '= (2 · ρ3 + 3 · p2 + pl + p0 + q0 + 4) >> 3 · · · (21) [0129] When ChiOmaEdgeFlag indication "O" does not satisfy the equation (18) conditions, the deblocking filter according to equation (22) IlJ (24) calculate pixel data pO ', PI' and p2 '. [0130] pO' = (2 · pl+p 0+ql+2) >> 2 · · · (22)[0131] ρΓ = pi · · · (23)[0132] ρ2' = ρ2 · · · (24)[0133] 当ChiOmaEdgeFlag指示“O”且满足公式(25)的条件时,去块滤波器根据公式(26) 到(28)计算像素数据qO,、ql,和q2,。 [0130] pO '= (2 · pl + p 0 + ql + 2) >> 2 · · · (22) [0131] ρΓ = pi · · · (23) [0132] ρ2' = ρ2 · · · ( 24) [0133] When ChiOmaEdgeFlag indication "O" and satisfy the following formula (25) conditions, the deblocking filter (26) to (28) the pixel data calculated according to the formula qO ,, ql, and q2 ,. [0134] aq < β && | p0-q0 < ((α >> 2)+2) · · · (25)[0135] qO' = (pl+2 · ρΟ+2 · qO+2 · ql+q2+4) >> 3 · · · (26)[0136] ql' = (p0+q0+ql+q2+2) >>2 · · · (27)[0137] q2' = (2 · q3+3 · q2+ql+q0+p4+4) >>3 · · · (28)[0138] 当ChiOmaEdgeFlag指示“0”且不满足公式(25)的条件时,去块滤波器根据公式(29)到(31)计算像素数据qO,、ql,和q2,。 [0134] aq <β && | p0-q0 <((α >> 2) +2) · · · (25) [0135] qO '= (pl + 2 · ρΟ + 2 · qO + 2 · ql + q2 +4) >> 3 · · · (26) [0136] ql '= (p0 + q0 + ql + q2 + 2) >> 2 · · · (27) [0137] q2' = (2 · q3 + 3 · q2 + ql + q0 + p4 + 4) >> 3 · · · (28) [0138] When ChiOmaEdgeFlag indicates "0" and does not satisfy the equation (25) the conditions of the deblocking filter in accordance with equation (29) to (31) calculates the pixel data qO ,, ql, and q2 ,. [0139] qO,= (2 · ql+qO+pl+2) >> 2 · · · (29)[0140] ql, = ql · · · (30)[0141] q2, = q2 · · · (31)[0142] 从而,在H264. /AVC的编码系统中,通过利用像素数据pO到p3和qO到q3进行滤波来计算像素数据PO'到p2'和qO'到q2'。 [0139] qO, = (2 · ql + qO + pl + 2) >> 2 · · · (29) [0140] ql, = ql · · · (30) [0141] q2, = q2 · · · ( 31) [0142] Thus, in H264. / AVC encoding system, by using the pixel data pO to p3 and q3 qO to be calculated by filtering the pixel data PO 'to p2' and qO 'to q2'. [0143] 〈3.图像编码装置中的去块滤波器和滤波器设置部分的配置〉[0144] 滤波器设置部分41根据关注宏块中的最优模式的预测块尺寸设置去块滤波器24 中的抽头长度和滤波对象像素范围。 [0143] <3 image encoding device deblocking filter and filter settings section of Configuration> [0144] filter setting section 41 to set the deblocking filter 24 according to the prediction block size concern macroblock optimal mode The tap length and the filter object pixel range. [0145] 通常,在较大块尺寸的情况中,块失真对于人眼是明显的。 [0145] In general, in the case of a large block size, block distortion is apparent to the human eye. 另外,对于不包括很多纹理信息的平坦区域倾向于选择较大块尺寸。 In addition, the flat area that does not include a lot of texture information tend to choose large block sizes. [0146] 从而,滤波器设置部分41根据彼此相邻的两个块的相邻边上的块尺寸进行情况分类,并且根据情况分类的结果设置滤波的抽头长度和滤波对象像素范围。 [0146] Thus, the filter setting section 41 according to the classification block size of two adjacent blocks adjacent to each other on the edge, and setting the tap length of the filter and the filtered target pixel range according to the classification results. 滤波器设置部分41进行情况分类为例如其中彼此相邻的两个块的相邻边都具有预定块尺寸或更小的块尺寸的情况、和其中彼此相邻的两个块中的至少一个具有大于预定块尺寸的扩展块尺寸的情况。 A filter setting section 41, for example, where the classification of two adjacent blocks adjacent to each other side block having a predetermined size or smaller block size case, adjacent to each other and wherein at least one of the two blocks having block size is larger than the predetermined extension block size. [0147] 在其中彼此相邻的两个块的相邻边上的块尺寸都具有预定块尺寸或更小的块尺寸的情况中,例如H. 264/AVC标准的宏块尺寸,滤波器设置部分41通过如上所述进行滤波来计算像素数据PO'到p2'和qO'到q2'。 Case [0147] in which the block size of the adjacent two blocks adjacent to each other on the edge of a block having a predetermined size or smaller block size, such as H. 264 / AVC standard macro-block size, filter settings section 41 calculates pixel data PO 'to p2' and qO 'to q2' is filtered as described above. 在其中彼此相邻的两个块中的至少一个具有大于预定块尺寸的扩展块尺寸的情况中,滤波器设置部分41根据块边界的块尺寸扩展抽头长度或滤波对象像素范围。 In the case where the two blocks adjacent to each other in at least one block having a predetermined size larger than the size of the extended block, the filter setting section 41 according to the block size of the block boundary extended target pixel filter tap length or range. 通过扩展抽头长度或滤波对象像素范围,滤波器设置部分41对具有明显块失真的较大块尺寸的部分进行更高强度的平滑处理和施加到离块边界更远的像素的值的滤波。 By extending the target pixel filter tap length or range, a filter setting section portion 41 having significantly larger block distortion smoothing block size and a higher strength value is applied to the filter farther away from the block boundary pixels. 从而,使得块失真不明显,并且使解码图像的主观图像质量更符合期望。 Thus, making the block distortion is not obvious, and the decoded image of the subjective image quality is more in line with expectations. [0148] 另外,当进行更高强度的平滑处理时,失去了图像的高频分量。 [0148] In addition, when smoothed higher intensity, it lost high-frequency components of the image. 然而,经常对具有小的高频分量的相对平坦的区域应用大的块尺寸,从而不出现诸如纹理遗失等的主观恶化。 Often, however, a small high-frequency component having a relatively flat area application large block size, so as not to appear as missing and other textures subjective worse. [0149] 滤波器设置部分41从而根据预测块尺寸生成指示抽头长度和滤波对象像素范围的参数值,并将参数值提供到去块滤波器24。 [0149] filter setting section 41 so as to generate parameter value indicates the tap length and the filter object pixel block size range according to forecasts, will provide parameter values to the deblocking filter 24. 另外,当使用大于预定宏块的块尺寸的多个不同尺寸的宏块时,随着块尺寸增大,滤波器设置部分41可以设置更长的抽头长度和更大的滤波对象像素范围。 Further, when a plurality of different sizes larger than the predetermined macroblock macroblock block size, with the block size increases, the filter setting section 41 may set a longer tap length and larger filter object pixel range. [0150] 图4示出去块滤波器和滤波器设置部分的配置。 [0150] FIG. 4 shows a block filter out part of the configuration and filter settings. 滤波器设置部分41包括块尺寸缓冲器411和参数值生成部分412。 Filter setting section 41 block size buffers 411 and parameter values include generating section 412. 另外,去块滤波器24包括滤波强度确定部分241和滤波部分242。 Further, the deblocking filter 24 comprises a filter strength determining section 241 and filter section 242. [0151] 块尺寸缓冲器411累积用于一个帧图像的信息,该信息指示在由预测图像和最优模式选择部分33选择的最优模式中的预测块尺寸。 [0151] block size for a frame buffer 411 cumulative image information that indicates the prediction image and optimum mode selecting section 33 selects the optimal prediction mode block sizes. 即,块尺寸缓冲器411存储关于一个帧图像中作为编码对象的每个宏块的预测块尺寸的信息。 That is, the buffer 411 stores the block size on a frame image information of each macroblock as the prediction of the encoding target block size. [0152] 参数值生成部分412基于块尺寸缓冲器411的预测块尺寸信息确定在彼此相邻的两个块的相邻边上的预测块尺寸。 [0152] Parameter value generation section 412 based on the size of the buffer block prediction block size information 411 to determine the two adjacent blocks adjacent to each other on the edge of the prediction block size. 参数值生成部分412基于确定的预测块尺寸生成用于对两个块之间的块边界设置滤波的抽头长度和滤波对象像素范围的参数值,并将该参数值提供到滤波强度确定部分241。 Parameter generating section 412 generates the tap length of the block boundary and the parameter values set filter between the two blocks of filtering target pixel based on the determined range of the prediction block size, and supplies the values to the filter strength parameter determining section 241. [0153] 滤波强度确定部分241基于从无损编码部分16提供的预测模式信息确定块边界强度数据Bs,并将确定的块边界强度数据Bs和从参数值生成部分412提供的参数值输出到滤波部分242。 [0153] filtering strength determination part 241 based on the prediction mode information from the lossless encoding section 16 provides determination block boundary strength data Bs, and determining a block boundary strength data Bs and the parameter value generation section 412 supplied from the parameter value is output to the filter section 242. [0154] 滤波部分242利用块边界强度数据Bs以及由参数值指示的抽头长度和滤波对象像素范围进行滤波以计算滤波后的像素数据。 [0154] filtering section 242 use the block boundary strength data Bs and tap length and the filter object pixel value indicated by the scope of filtering parameters to calculate the filtered pixel data. [0155] 下面将示出分类为如下情况的情况分类:第一情况,其中彼此相邻的两个块的相邻边的预测块尺寸都是预定块尺寸(16 X 16像素)或更小的块尺寸;以及第二情况,其中彼此相邻的两个块中的至少一个具有大于预定块尺寸的扩展块尺寸。 Will be shown [0155] The following is a case of the classification category: a first case in which the adjacent edges of each adjacent two blocks of the prediction block size is a predetermined block size (16 X 16 pixels) or smaller block size; and a second case in which two blocks adjacent to each other in at least one block having a predetermined size larger than the extension block size. 在该情况中,在第一情况中进行上述H. 264/AVC编码系统的滤波。 In this case, filtering said H. 264 / AVC encoding system in the first case. 另外,在第二情况中,将抽头长度设置为扩展的长度以增大平滑强度,和/或扩展滤波对象像素范围以对远离块边界的位置处的像素进行滤波。 In the second case, the tap length is set to expand in order to increase the length of the smooth strength, and / or expanded range filter object pixel pixel at a position away from the block boundary are filtered. [0156] 下面将描述当扩展抽头长度和滤波对象像素范围时的滤波。 [0156] The following description will be extended the length of filter taps and the filter object pixel range. [0157] 滤波部分242基于参数值扩展抽头长度和滤波对象像素范围,进行滤波,并根据像素数据PO到p4和qO到q4计算滤波后的像素数据pO'到p3,和qO'到q3,。 [0157] filtering section 242 based on the parameter spread tap length and the filter object pixel range, filtering, and based on the pixel data to the pixel data PO pO p4 and qO to calculate filtered q4 'to p3, and qO' to q3 ,. 在该情况中,滤波部分242使用公式(32 )代替上述公式(7 )。 In this case, filtering section 242 using equation (32) instead of the equation (7). [0158] Δ =Clip3(_tc, tc ((((q0-p0) < < 3) + ((pl-ql) < < I) + (p2_q2)+8) > > 4))• · · (32)[0159] 另外,滤波部分242通过使用公式(33 )和(34 )代替公式(12 )和(14 )计算像素数据ρΓ和ql'。 [0158] Δ = Clip3 (_tc, tc ((((q0-p0) <<3) + ((pl-ql) <<I) + (p2_q2) +8)>> 4)) • · · (32 ) [0159] In addition, the filter section 242 by using the formula (33) and (34) instead of the formula (12) and (14) to calculate the pixel data ρΓ and ql '. [0160] pi,= pl+Clip3 (_tc0, tcO, (p3+p2+p0+ ((q0+ql+l) >> I) -(p I << 2)) >> 2) · · · (33)[0161] ql,= ql+Clip3 (_tc0, tcO, (q3+q2+q0+ ((q0+ql+l) > > I) - (ql < < 2)) > > 2)• · · (34) [0162] 另外,滤波部分242通过使用公式(35)和(36)代替公式(16)和(17)计算像素数 据P2'和q2' O[0163] P2, =p2+Clip3 (_tc0,tcO,(p4+p3+pl+ ((ρθ+ql+l) >> l)-(p2 << 2)) >>2)• · · (35) [0164] q2, =q2+Clip3 (_tc0,tcO,(q4+q3+ql+((q0+ql+l) >> l)-(q2 << 2)) >>2)• · · (36) [0165] 另外,当chromaEdgeFlag指示“O”且满足公式(18)的条件时,滤波部分242根据 公式(37 )到(40 )计算像素数据pO '、P1'、p2 '和p3 '。 [0160] pi, = pl + Clip3 (_tc0, tcO, (p3 + p2 + p0 + ((q0 + ql + l) >> I) - (p I << 2)) >> 2) · · · (33 ) [0161] ql, = ql + Clip3 (_tc0, tcO, (q3 + q2 + q0 + ((q0 + ql + l)>> I) - (ql <<2))>> 2) • · · (34 ) [0162] In addition, the filtering section 242 (35) and (36) instead of the equation by using the formula (16) and (17) calculating the pixel data P2 ', and q2' O [0163] P2, = p2 + Clip3 (_tc0, tcO , (p4 + p3 + pl + ((ρθ + ql + l) >> l) - (p2 << 2)) >> 2) • · · (35) [0164] q2, = q2 + Clip3 (_tc0, tcO , (q4 + q3 + ql + ((q0 + ql + l) >> l) - (q2 << 2)) >> 2) • · · (36) [0165] In addition, when chromaEdgeFlag indication "O" and satisfies When the equation (18) condition, the filtering section 242 according to equation (37) to (40) calculating the pixel data pO ', P1', p2 'and p3'. [0166] PO, =(p3+2 · p2+3 · pl+4 · pO+3 · qO+2 · ql+q2+8) >>4 · · · (37)[0167] Pi, =(p3+p2+2 · pl+2 · pO+qO+ql+4) >>3[0168] • · · (38)[0169] P2, =(p4+3 · p3+4 · p2+3 · pl+2 · pO+2 · qO+ql+8) >>4 · · · (39)[0170] p3, =(p4+3 · p3+p2+pl+p0+q0+4) >>3[0171] • · · (40)[0172] 另外,当chromaEdgeFlag指示“O”且满足公式(25)的条件时,滤波部分242根据 公式(41)到(44)计算像素数据qO,、ql,、q2,和q3,。 [0166] PO, = (p3 + 2 · p2 + 3 · pl + 4 · pO + 3 · qO + 2 · ql + q2 + 8) >> 4 · · · (37) [0167] Pi, = (p3 + p2 + 2 · pl + 2 · pO + qO + ql + 4) >> 3 [0168] • · · (38) [0169] P2, = (p4 + 3 · p3 + 4 · p2 + 3 · pl + 2 · pO + 2 · qO + ql + 8) >> 4 · · · (39) [0170] p3, = (p4 + 3 · p3 + p2 + pl + p0 + q0 + 4) >> 3 [0171] When • · · (40) [0172] In addition, when chromaEdgeFlag indication "O" and satisfy the following formula (25) condition, the filtering section 242 according to equation (41) to (44) calculating the pixel data qO ,, ql ,, q2, and q3 ,. [0173] qO, =(p2+2 · pl+3 · pO+4 · qO+3 · ql+2 · q2+q3+8) >>4 · · · (41)[0174] ql, =(pl+pO+2 · qO+2 · ql+q2+q3+4) >>3[0175] • · · (42)[0176] q2, =(q4+3 · q3+4 · q2+3 · ql+2 · qO+2 ·pO+pl+8) >>4 · · · (43)[0177] q3, =(q4+3 · q3+q2+ql+q0+p0+4) >>3[0178] • · · (44)[0179] 另外,对抽头长度和滤波对象像素范围的设置不限于分类为两种情况(即,其中 两个块都具有16X16像素的尺寸或更小的尺寸的情况和其中两个块中的至少一个大于16X16像素的情况)的情况分类。 [0173] qO, = (p2 + 2 · pl + 3 · pO + 4 · qO + 3 · ql + 2 · q2 + q3 + 8) >> 4 · · · (41) [0174] ql, = (pl + pO + 2 · qO + 2 · ql + q2 + q3 + 4) >> 3 [0175] • · · (42) [0176] q2, = (q4 + 3 · q3 + 4 · q2 + 3 · ql + 2 · qO + 2 · pO + pl + 8) >> 4 · · · (43) [0177] q3, = (q4 + 3 · q3 + q2 + ql + q0 + p0 + 4) >> 3 [0178] • · · (44) [0179] In addition, setting the tap length of the filter and is not limited to the range of the target pixel classified into two cases (i.e., both of which have a 16X16 pixel block size or smaller size of the case and wherein in the case of at least one of the two blocks is larger than the case of 16X16 pixels) classification. 例如,可以进行情况分类,以分类为例如其中两个块都具有16X16像素的尺寸或更小尺寸的情况、其中两个块中的至少一个大于16X 16像素且两个块都为32 X 32像素或更小的情况、以及其中两个块的至少一个大于32X32像素的情况。 For example, the classification can be carried out, for example, to classify the two blocks have a situation wherein the 16X16 pixel size or smaller size, wherein at least one of the two blocks is greater than 16X 16 pixels and two blocks of 32 X 32 pixels or less of the case, and wherein at least one of the two blocks is greater than 32X32 pixels. 在该情况中,在较大块尺寸的边界处,进一步加长抽头长度以增加平滑强度,并且进一步增大滤波对象像素范围以对远离块边界的像素值进行滤波。 In this case, at the boundary of the larger block size, in order to further increase tap length longer smooth strength, and further increase the range of the filter away from the target pixel block boundary filtering pixel values. 另外,在设置抽头长度和滤波对象像素范围中,根据情况分类的结果可以仅增大抽头长度和滤波对象像素范围之一。 In addition, setting the tap length and the filter object pixel range, depending on the classification of the results can only increase the length of the filter tap one of the objects, and pixel bounds. [0180] 从而,图像编码装置根据在彼此相邻的两个块的相邻边上的块尺寸设置滤波器的抽头长度和滤波对象像素范围,并对具有明显块失真的较大块尺寸的部分进行更高强度的平滑处理和应用于更远离块边界的像素的值的滤波。 Part [0180] Thus, the image coding apparatus according to filter tap length and the range of the target pixel in two adjacent blocks adjacent to each other on the edge of the block size setting of the filter, and block distortion obviously larger block size smoothing applied higher strength and more away from the block boundary filtering values of pixels. 从而,使得块失真不明显,并且可以使得用于生成预测图像的解码图像的图像质量更符合期望。 Thereby, making the block distortion inconspicuous, and image quality can be made for the predicted image generating decoded image more in line with expectations. [0181] <4.图像编码装置的操作〉[0182] 下面将描述图像编码处理操作。 [0181] <4 image coding device operation> [0182] The following image coding process will be described. 图5示出用于图像编码处理的预测块尺寸。 Figure 5 shows a prediction block size used in image coding processing. H. 264/AVC系统定义如图5 (C)和图5 (D)所示的16X 16像素和4X4像素的预测块尺寸。 H. 264 / AVC system defined in Figure 5 (C) and Figure 5 (D) 16X 16 pixel and the prediction block size 4X4 pixel shown. 另外,当使用大于H. 264/AVC系统的尺寸的扩展尺寸的宏块时,例如,当使用32X32像素的宏块时,定义例如如图5 (B)所示的预测块尺寸。 In addition, when using more than H. 264 / AVC system, the size of the expansion of the macro block size, for example, when using a 32X32 pixel macro block is defined for example shown in Figure 5 (B) of the prediction block size. 当使用64X64像素的宏块时,定义例如图5 (A)所示的预测块尺寸。 When using a 64X64 pixel macro block is defined for example in Fig. 5 (A) shown in the prediction block size. [0183] 另外,图5中的“跳跃/直接”表示当在运动预测和补偿部分32中选择跳跃宏块或直接模式时的预测块尺寸。 [0183] In FIG. 5, "skip / direct" means that when the motion prediction and compensation portion 32 select skip or direct mode macroblock prediction block size. 另外,“ME”表示运动补偿块尺寸。 Further, "ME" indicates motion compensation block size. 另外,“P8X8”表示可以在减小宏块的尺寸的更低层中进一步划分。 Further, "P8X8" represents a lower layer further reduce the size of divided macroblock. [0184] 图6为图像编码处理操作的流程图。 [0184] FIG. 6 is a flowchart of the operation of the image coding process. 在步骤ST11,A/D转换部分11对输入图像信号进行Α/D转换。 In step ST11, A / D conversion section 11 of the input image signal Α / D conversion. [0185] 在步骤ST12,画面重排列缓冲器12进行画面重排列。 [0185] In step ST12, the picture rearrangement buffer 12 picture rearrangement. 画面重排列缓冲器12存储从Α/D转换部分11提供的图像数据,并将显示每个画面的顺序重排列为编码每个画面的顺序。 Picture rearrangement buffer 12 storing section 11 converts the image data supplied from Α / D, and the display order of each picture is rearranged sequences encoding each picture. [0186] 在步骤ST13,减法部分13生成预测误差数据。 [0186] In step ST13, the subtraction part 13 generates a prediction error data. 减法部分13通过计算在步骤ST12 中重排列的图像的图像数据与通过预测图像和最优模式选择部分33选择的预测图像数据之差而生成预测误差数据。 Subtraction section 13 and the selecting section 33 selects the optimum mode by predicting the difference between the prediction image and the image data of the prediction error is generated by the image data of the image data to calculate rearranged in step ST12. 相比于原始图像数据,该预测误差数据在量上减小。 Compared to the original image data, the prediction error is reduced in the amount of data. 从而,相比于按原样对图像进行编码的情况,能够压缩数据量。 Thus, the circumstances as compared to the image coding can compress the data. 另外,当预测图像和最优模式选择部分33以片段为单位选择从帧内预测部分31提供的预测图像和来自运动预测和补偿部分32的预测图像时,在选择了从帧内预测部分31提供的预测图像的片段中进行帧内预测。 Further, when a prediction image and optimum mode selecting section 33 units predicted image selecting section 31 and the predicted image supplied from the motion prediction and compensation portion 32 with the prediction from the intra-fragment, selected from the intra prediction section 31 provides The predicted fragment of an image intra prediction. 另外, 在选择了来自运动预测和补偿部分32的预测图像的片段中进行帧间预测。 Further, the predicted image segment is selected from the motion prediction and compensation portion 32 of interframe prediction. [0187] 在步骤ST14,正交变换部分14进行正交变换处理。 [0187] In step ST14, the orthogonal transform section 14 performs orthogonal transform processing. 正交变换部分14对从减法部分13提供的预测误差数据进行正交变换。 Orthogonal transform section 14 the prediction error data supplied from the subtraction portion 13 performs orthogonal transform. 具体是,对预测误差数据进行诸如离散余弦变换、Karhunen-Loeve变换等的正交变换,并输出变换系数数据。 More specifically, the prediction error data such as discrete cosine transform, Karhunen-Loeve transform of the orthogonal transform, and outputs transform coefficient data. [0188] 在步骤ST15,量化部分15进行量化处理。 [0188] In step ST15, the quantization section 15 performs quantization processing. 量化部分15量化变换系数数据。 Quantization section 15 quantizes the transform coefficient data. 在量化中,如下文将对步骤ST25的处理中的描述,进行速率控制。 In the quantization, as described below will be treated as described in step ST25, and perform rate control. [0189] 在步骤ST16,去量化部分21进行去量化处理。 [0189] In step ST16, the dequantization section 21 to perform quantization processing. 去量化部分21用与量化部分15的特征对应的特征对通过量化部分15量化的变换系数数据进行去量化。 Dequantizing portion 21 and the quantization portion 15 with the characteristic features of the corresponding quantization section 15 by the quantized transform coefficients dequantized data. [0190] 在步骤ST17,反向正交变换部分22进行反向正交变换处理。 [0190] In step ST17, inverse orthogonal transform section 22 performs inverse orthogonal transform processing. 反向正交变换部分22以对应于正交变换部分14的特征的特征对通过去量化部分21去量化的变换系数数据进行反向正交变换。 Inverse orthogonal transform portion 22 corresponding to the orthogonal transform portion 14 characterized by the features of the dequantizing quantized transform portion 21 to inverse orthogonal transform coefficient data. [0191] 在步骤ST18,加法部分23生成解码图像数据。 [0191] In step ST18, the adding section 23 generates decoded image data. 加法部分23通过将从预测图像和最优模式选择部分33提供的预测图像数据与位于对应于预测图像的位置的经过反向正交变换后的数据加到一起而生成解码图像数据。 Data from the addition section 23 by the optimum prediction image and the prediction mode selecting section 33 provides image data corresponding to the location in the predicted image through the inverse orthogonal transform are added together to generate decoded image data. [0192] 在步骤ST19,去块滤波器24进行滤波。 [0192] In step ST19, the deblocking filter 24 for filtering. 去块滤波器24通过对从加法部分23输出的解码图像数据进行滤波去除块失真。 Deblocking filter 24 through the decoded image data output from the adding section 23 is filtered to remove the block distortion. [0193] 在步骤ST20,帧存储器25存储解码图像数据。 [0193] In step ST20, the frame memory 25 stores the decoded image data. 帧存储器25存储滤波前的解码图像数据和滤波后的解码图像数据。 Decoded image data decoded image data and the filtered frame memory 25 stores before filtering. [0194] 在步骤ST21,帧内预测部分31和运动预测和补偿部分32各自进行预测处理。 [0194] In step ST21, the intra prediction section 31 and the motion prediction and compensation portion 32 each prediction processing. 具体是,帧内预测部分31在帧内预测模式中进行帧内预测处理,以及运动预测和补偿部分32 在帧间预测模式中进行运动预测和补偿处理。 Specifically, the intra prediction section 31 performs intra prediction processing in the intra prediction mode, and motion prediction and compensation portion 32 performs motion prediction and compensation processing in the inter prediction mode. 下文将参考图7详细描述预测处理的内容。 Will hereinafter be described in detail with reference to FIG content prediction process. 该处理在作为候选的全部预测模式中进行对应的预测处理,并计算作为候选的全部预测模式的对应的成本函数值。 The process proceeds in a prediction process corresponding to all the candidate prediction modes, and the cost function value calculated as corresponding to all the candidate prediction modes. 然后,基于计算的成本函数值选择最优帧内预测模式和最优帧间预测模式,并将在选择的预测模式中生成的预测图像、其成本函数、以及预测模式信息提供到预测图像和最优模式选择部分33。 Then, based on the calculated cost function values to select the optimal intra prediction mode and the optimal inter prediction mode, and generated in the selected prediction mode, a prediction image, its cost function, and is supplied to the prediction mode information and the prediction image Excellent mode selection section 33. [0195] 在步骤ST22,预测图像和最优模式选择部分33选择预测图像数据。 [0195] In step ST22, the optimum prediction image and the prediction mode selecting section 33 selects the image data. 预测图像和最优模式选择部分33基于从帧内预测部分31和运动预测和补偿部分32输出的对应的成本函数值确定其中获得最佳编码效率的最优模式。 Prediction image and optimum mode selecting section 33 based on the prediction and compensation portion 32 corresponding to the output from the intra prediction section 31 and the motion of the cost function value determines the best mode in which the optimal encoding efficiency. 另外,预测图像和最优模式选择部分33在确定的最优模式中选择预测图像数据,并将预测图像数据提供到减法部分13和加法部分23。 Further, the optimal mode prediction image and the predicted image selecting section 33 selects the optimum mode is determined in the data, and the predictive image data supplied to the subtraction section 13 and addition section 23. 如上所述,在步骤ST13和ST18的操作中使用该预测图像。 As described above, the use of the prediction image in step ST13 and ST18 operations. 另外,将对应于选择的预测图像数据的预测模式信息输出到无损编码部分16和滤波器设置部分41。 Further, the prediction mode corresponding to the selected prediction image data information output to the lossless encoding section setting section 16 and the filter 41. [0196] 在步骤ST23,无损编码部分16进行无损编码处理。 [0196] In step ST23, the lossless encoding section 16 performs lossless coding process. 无损编码部分16对从量化部分15输出的量化数据进行无损编码。 Lossless coding section 16 quantized data output from the quantization section 15 lossless encoding. 具体是,对量化数据进行诸如可变长度编码、算术编码等的无损编码,并压缩数据。 More specifically, the quantized data such as variable length coding, lossless coding such as arithmetic coding, and compressed data. 此时,对在上述步骤ST22等中输入到无损编码部分16的预测模式信息(包括例如宏块类型、预测模式、运动矢量信息、参考画面信息等)也进行无损编码。 At this time, like in the above-described step ST22 is input to the lossless encoding prediction mode information section 16 (e.g., including macro-block type, prediction mode, motion vector information, reference picture information, etc.) is also lossless encoding. 另外,将预测模式信息的无损编码数据添加到通过无损编码量化数据生成的编码流的头信息中。 Also, add lossless encoding data prediction mode information to the header information generated by lossless encoding quantized data encoded stream. [0197] 在步骤ST24,存储缓冲器17进行存储处理以存储编码流。 [0197] In step ST24, the storage buffer 17 to store the encoded stream storage process. 适当时读取存储在存储缓冲器17中的编码流,并经传输线将其传输到解码侧。 Appropriate reading stored in the memory buffer 17 coded stream, and its transmission via the transmission line to the decoding side. [0198] 在步骤ST25,速率控制部分18进行速率控制。 [0198] In step ST25, the rate control section 18 rate control. 速率控制部分18控制量化部分15 的量化操作的速率,以在存储缓冲器17存储编码流时防止存储缓冲器17上溢或下溢。 Rate control section 18 controls the quantization rate operation part 15 in order to prevent the storage buffer when the storage buffer 17 stores the coded stream 17 overflow or underflow. [0199] 下面将参考图7的流程图描述图6中步骤ST21中的预测处理。 [0199] The following will describe the flowchart of FIG. 7 6 steps ST21 prediction process with reference to FIG. [0200] 在步骤ST31,帧内预测部分31进行帧内预测处理。 [0200] In step ST31, the intra prediction section 31 performs intra prediction process. 帧内预测部分31在作为候选的全部帧内预测模式中对作为处理对象的块的图像进行帧内预测。 Intra prediction section 31 as a whole in the intra prediction mode candidate in the image processing target block is intra prediction. 另外,将未通过去块滤波器24滤波的存储在帧存储器25中的解码图像数据用作在帧内预测中引用的解码图像的图像数据。 Moreover, the deblocking filter 24 is stored in the frame memory 25 filtering the decoded image data as reference image data in the intra-prediction decoded image will not pass. 将在下文中详细描述帧内预测处理的内容。 The detailed description of the contents of the intra prediction processing hereinafter. 该处理在作为候选的全部帧内预测模式中进行帧内预测处理,并计算作为候选的全部帧内预测模式的成本函数值。 The process proceeds as a candidate in the intra prediction processing of all the intra prediction mode and calculates the cost value as a function of all the intra-prediction mode candidate. 然后,基于计算的成本函数值从全部帧内预测模式中选择获得最佳编码效率的一个帧内预测模式。 Then, based on the calculated cost function values to select an optimum encoding efficiency from all the intra-prediction mode in intra prediction mode. [0201 ] 在步骤ST32,运动预测和补偿部分32进行帧间预测处理。 [0201] In step ST32, the motion prediction and compensation portion 32 performs inter-frame prediction processing. 运动预测和补偿部分32 使用存储在帧存储器25中的滤波后的解码图像数据在作为候选的全部帧间预测模式(全部预测块尺寸)中进行帧间预测处理。 Decoded image data and the motion compensation prediction section 32 stored in the frame memory 25 after filtering interframe prediction process as a whole in the inter prediction mode candidate (all prediction block size) in. 将在下文中详细描述帧间预测处理的内容。 Content interframe prediction process will be described in detail hereinafter. 该处理在作为候选的全部帧间预测模式中进行帧间预测处理,并计算作为候选的全部帧间预测模式的成本函数值。 The process proceeds interframe prediction process as a candidate in all the inter prediction mode and calculates the cost value as a function of all the inter prediction mode candidate. 然后,基于计算的成本函数值从全部帧间预测模式中选择获得最佳编码效率的一个帧间预测模式。 Then, based on the calculated cost function values to select an optimum encoding efficiency from all the inter prediction modes in inter prediction mode. [0202] 下面将参考图8的流程图描述图7中步骤ST31中的帧内预测处理。 [0202] described below with reference to a flowchart of FIG. 7, FIG. 8 intra prediction processing step of ST31. [0203] 在步骤ST41,帧内预测部分31在每个预测模式中进行帧内预测。 [0203] In step ST41, the intra prediction section 31 performs intraframe prediction in each prediction mode. 帧内预测部分31利用存储在帧存储器25中的滤波前的解码图像数据对于每个帧内预测模式生成预测图像数据。 Intra prediction predictive image data generated using a memory section 31 for each intra-prediction mode decoded image data in the frame memory 25 before the filtering. [0204] 在步骤ST42,帧内预测部分31对于每个预测模式计算成本函数值。 [0204] In step ST42, the intra prediction section 31 for each prediction mode to calculate the cost function value. 基于如在H. 264/AVC系统中作为参考软件的JM (联合模型)中定义的高复杂性模式和低复杂性模式之一的方法计算成本函数值。 (Joint Model) defined in highly complex models and methods of low complexity mode function value is calculated based on the cost as in H. 264 / AVC system as a reference software JM. [0205] 具体是,在高复杂性模式中,如同步骤ST41的处理,对作为候选的全部预测模式试验性地进行无损编码处理,并且对于每个预测模式计算由下面的公式(45)表达的成本函数值。 [0205] Specifically, in the high complexity mode, as processing of step ST41, all of the candidate prediction modes as a trial carried out lossless encoding processing, and a calculation for each prediction mode (45) is expressed by the following formula cost function values. [0206]成本(模式e Ω) = ϋ+λ .R …(45)[0207] Ω表示用于编码关注块或宏块的作为候选的预测模式的全集。 [0206] cost (mode e Ω) = ϋ + λ .R ... (45) [0207] Ω expressed concern for coding blocks or as a candidate of the prediction mode macroblock Collection. D表示当在预测模式中进行编码时在解码图像与输入图像之间的差分能量(失真XR是生成的编码的量,包括正交变换系数、预测模式信息等。λ是根据量化参数QP给定的Lagrange乘子。[0208] S卩,在高复杂性模式中编码要求在作为候选的全部预测模式中进行一次试验性编码处理以计算上述参数D和R,并需要较大的操作量。[0209] 另一方面,在低复杂性模式中,如同步骤ST41中的处理,生成预测图像,并且对于作为候选的全部预测模式计算达到头比特的运动矢量信息、预测模式信息等,并且对于每个预测模式计算通过下面的公式(46)表达的成本函数值。[0210]成本(模式 G Ω ) = D+QPtoQuant (QP) · header_bit ... (46)[0211] Ω表示用于编码关注块或宏块的作为候选的预测模式的全集。D表示当在预测模式中进行编码时在解码图像与输入图像之间的差分能量(失真XheaderjDit是用于预测模式的头比特。QPtoQuant是根据量化参数QP给定的函数。[0212] S卩,低复杂性模式要求对于每个预测模式进行预测处理,但是不需要解码图像,从而能够以小于高复杂性模式的操作量实现预测处理。[0213] 在步骤ST43,帧内预测部分31确定最优帧内预测模式。基于在步骤ST42中计算的成本函数值,帧内预测部分31选择其成本函数值是各个成本函数值中的最小值的一个帧内预测模式,并确定该帧内预测模式作为最优帧内预测模式。[0214] 下面将参考图9的流程图描述图7中步骤ST32中的帧间预测处理。[0215] 在步骤ST51,运动预测和补偿部分32对于每个预测模式确定运动矢量和参考图像。即,运动预测和补偿部分32在每个预测模式中对于作为处理对象的块确定运动矢量和参考图像。[0216] 在步骤ST52,运动预测和补偿部分32对于每个预测模式进行运动补偿。运动预测和补偿部分32基于在步骤ST51中确定的运动矢量在每个预测模式中(每个预测块尺寸) 对参考图像施加运动补偿,并对每个预测模式生成预测图像数据。[0217] 在步骤ST53,运动预测和补偿部分32对于每个预测模式生成运动矢量信息。运动预测和补偿部分32生成将被包括在编码流中的运动矢量信息,所述信息是关于在每个预测模式中确定的运动矢量。例如,利用中间预测等确定预测运动矢量,并生成指示通过运动预测检测的运动矢量与预测运动矢量之差的运动矢量信息。 When D represents the prediction encoding mode energy difference between the decoded image and the input image (the amount of coding distortion XR is generated, including orthogonal transform coefficient, prediction mode information, etc. .λ is given according to the quantization parameter QP Lagrange multiplier. [0208] S Jie, in high complexity mode coding requirements for a pilot coding process to calculate the above parameters D and R, and requires a large amount of operation as a candidate in all prediction modes. [ 0209] On the other hand, the low complexity mode, as in the processing in step ST41, prediction image is generated, and for calculating a prediction mode candidate reaches all header bits of the motion vector information, prediction mode information, etc., and for each calculating the cost function value prediction mode by the following equation (46) Expression. [0210] Cost (Mode G Ω) = D + QPtoQuant (QP) · header_bit ... (46) [0211] Ω denotes a coded block of interest or as a candidate macroblock prediction mode indicates the first bit .QPtoQuant Collection .D when encoding the prediction mode in the energy difference between the decoded image and the input image (distortion XheaderjDit prediction mode is used based on the quantization parameter QP given function. [0212] S Jie, low complexity model requires that for each prediction mode prediction process, but does not need to decode the image, allowing less than the amount of high complexity mode operation to achieve the prediction process. [0213] In step ST43, the intra prediction section 31 determines the optimum intra prediction mode based on the cost function value calculated in step ST42, the intra prediction section 31 selects the value of the cost function is the cost function value of each frame in a minimum intra-prediction mode and intra-prediction mode is determined as the optimum intra-prediction mode. [0214] described below with reference to a flowchart of FIG. 7 in step ST32 the inter-frame prediction processing. [0215] In step ST51, motion prediction and compensation portion 32 determines a motion vector for each prediction mode and the reference image. That is, the motion prediction and compensation portion 32 in each prediction mode for a processing target block determines the motion vector and the reference image. [0216] In step ST52, motion prediction and compensation portion 32 for each prediction mode motion compensation prediction and motion compensation section 32 in step ST51 based on the determined motion vector (each prediction block size) is applied to the motion of the reference image in each prediction mode compensation, and each of the prediction modes to generate predicted image data. [0217] In step ST53, the motion prediction and compensation portion 32 generates motion vector information for each prediction mode and the motion compensation prediction section 32 generates will be included in the encoded stream The motion vector information, the information about the motion vectors determined in each prediction mode. For example, using the intermediate prediction motion vector prediction is determined, and generates the motion indicated by the motion vector prediction with the detected motion vector of the difference motion prediction vector information. 这样生成的运动矢量信息还在下一个步骤ST54中用于计算成本函数值,并被包括在预测模式信息中,并在预测图像和最优模式选择部分33最后选择对应的预测图像时被输出到无损编码部分16。 Motion vector information thus generated also at the next step ST54 for calculating the cost function values, and are included in the prediction mode information and the prediction image and optimum mode selecting section 33 to select the corresponding final prediction image is output to the lossless encoding section 16. [0218] 在步骤ST54,运动预测和补偿部分32对于每个帧间预测模式计算成本函数值。 [0218] In step ST54, the motion prediction and compensation portion 32 for each inter prediction mode to calculate the cost function value. 运动预测和补偿部分32利用上述公式(45)或公式(46)计算成本函数值。 Motion prediction and compensation portion 32 by the above formula (45) or equation (46) calculating the cost function value. 另外,对于帧间预测模式计算成本函数值包括评估在H. 264/AVC系统中定义的跳跃模式和直接模式中的成本函数值。 In addition, for the inter prediction mode to calculate the cost function values including the assessment of the cost function value in H. 264 / AVC system defined in skip mode and direct mode. [0219] 在步骤ST55,运动预测和补偿部分32确定最优帧间预测模式。 [0219] In step ST55, the motion prediction and compensation portion 32 determines the optimal inter prediction mode. 基于在步骤ST54 中计算的成本函数值,运动预测和补偿部分32选择其成本函数值是各成本函数值中的最小值的一个预测模式,并确定该预测模式作为最优帧间预测模式。 Based on the cost function value calculated in step ST54, the motion prediction and compensation portion 32 selects the cost of each cost function value is a function of the minimum value of a predictive model, and determining the optimal inter prediction mode as the prediction mode. [0220] 下面将参考图10的流程图描述滤波器设置处理。 [0220] be described below with reference to a flowchart of setting processing filter 10 of FIG. 另外,图10表示其中扩展抽头长度和滤波对象像素范围的情况。 Further, FIG. 10 shows the case where the extended target pixel and filter tap length range. [0221] 在步骤ST61,滤波器设置部分41获得最优模式中的预测块尺寸。 [0221] In step ST61, the filter setting section 41 to obtain the optimal mode prediction block size. 滤波器设置部分41获得对应于在图6的步骤ST22中选择的预测图像的预测块尺寸,即,当在最优模式中进行编码时的预测块尺寸。 A filter setting section 41 to obtain a corresponding prediction block size in FIG step ST22 6 selected prediction image, i.e., when encoding in the optimal mode prediction block size. [0222] 在步骤ST62,滤波器设置部分41确定关注块或相邻块是否大于16X16像素。 [0222] In step ST62, the filter setting section 41 determines whether the block of interest or adjacent blocks larger than 16X16 pixels. 当关注块和相邻块中至少一个大于16X16像素时,滤波器设置部分41前进到步骤ST63。 When the attention block and adjacent blocks in at least one of more than 16X16 pixels, the filter setting section 41 proceeds to step ST63. 当关注块和相邻块都是16X16像素或更小时,滤波器设置部分41前进到步骤ST64。 When the attention block and adjacent blocks are 16X16 pixels or more hours, the filter setting section 41 proceeds to step ST64. [0223] 在步骤ST63,滤波器设置部分41扩展且设置抽头长度和滤波对象像素范围。 [0223] In step ST63, filter settings, and set the tap portion 41 extended length and filtered target pixel range. 例如,滤波器设置部分41将抽头长度和滤波对象像素范围扩展为大于在H. 264/AVC编码系统中的值,并使得如上所述计算滤波后的像素数据PO'到p3'和qO'到q3'。 For example, the filter tap length setting portion 41 and filtering the target pixel is greater than the value in the extended range of H. 264 / AVC encoding system, and calculated as described above so that the filtered pixel data PO 'to p3' and qO 'to q3 '. [0224] 在步骤ST64,滤波器设置部分41不扩展地设置抽头长度和滤波对象像素范围。 [0224] In step ST64, the filter setting section 41 does not extend to setting the tap length and the filter object pixel range. 例如,滤波器设置部分41设置H. 264/AVC系统的抽头长度和滤波对象像素范围,并使得如上所述计算滤波后的像素数据PO'到p2'和qO'到q2'。 For example, the filter setting section 41 to set the target pixel filter tap length and scope of H. 264 / AVC system, and calculated as described above so that the pixel data is filtered PO 'to p2' and qO 'to q2'. [0225] 从而,根据应用本发明的图像编码装置和图像编码方法,确定提供最佳编码效率的预测块尺寸,并按照确定的预测块尺寸来编码图像数据。 [0225] Thus, according to the image coding apparatus and image coding method of the present invention, to provide the best coding efficiency is determined prediction block size, and in accordance with the determined prediction block size of the encoded image data. 此时,在滤波器设置部分41的块尺寸缓冲器411中存储指示预测块尺寸的信息。 In this case, the filter portion 41 of the block set size stored in the buffer 411 indicating the prediction block size information. 从而,当通过对按照提供最佳编码效率的预测块尺寸编码的图像数据进行解码生成解码图像数据时,解码图像中的预测块的位置是清晰的。 Thus, when the follow through to provide the best coding efficiency prediction block size of a coded image data decoding to generate decoded image data, the position prediction decoded image block is clear. 从而,基于存储在块尺寸缓冲器411中的信息,根据预测块尺寸设置抽头长度和滤波对象像素范围,从而即使预测块尺寸较大也可以减小块失真。 Thus, based on the information stored in the buffer 411 in block size, based on the prediction block size setting target pixel and filter tap length range, so even if the prediction block size of the block distortion can be reduced greatly. 另外,由于可以减小在用于生成预测图像的解码图像数据中的块失真,从而可以防止由于块失真效应导致的预测误差数据量的增加。 Further, since the decoded image data can be reduced in a prediction image is generated for the block distortion, thereby preventing an increase in the amount of data due to the prediction error block distortion effect caused. 从而,可以进一步减小编码处理后的数据量。 Thus, the amount of data can be further reduced after the coding process. [0226] <5.图像解码装置的配置〉[0227] 经预定传输线、记录介质等将通过对输入图像进行编码而生成的编码流提供到图像解码装置并解码。 [0226] <5. The image decoding apparatus Configuration> [0227] through a predetermined transmission line, the recording medium or the like by encoding the input image to generate encoded stream is supplied to the image decoding apparatus and the decoding. [0228] 图11示出图像解码装置的配置。 [0228] FIG. 11 illustrates a configuration of the image decoding apparatus. 图像解码装置50包括存储缓冲器51、无损解码部分52、去量化部分53、反向正交变换部分54、加法部分55、去块滤波器56、画面重排列缓冲器57、以及D/Α转换部分58。 The image decoding apparatus 50 includes a storage buffer 51, the lossless decoding section 52, dequantization section 53, an inverse orthogonal transformation section 54, addition section 55, a deblocking filter 56, the picture rearrangement buffer 57, and D / Α conversion section 58. 图像解码装置50还包括帧存储器61、选择器62和65、帧内预测部分63、运动补偿部分64、以及滤波器设置部分71。 The image decoding apparatus 50 further includes a frame memory 61, the selector 62 and 65, the intra prediction section 63, motion compensation section 64, and a filter setting section 71. [0229] 存储缓冲器51存储传输的编码流。 [0229] storage buffer 51 stores the encoded transport stream. 无损解码部分52通过对应于图I中的无损编码部分16的编码系统的系统对从存储缓冲器51提供的编码流进行解码。 Lossless decoding section 52 coding system system portion 16 provided from the storage buffer 51 coded stream decoded by corresponding to Fig. I in lossless coding. 另外,无损解码部分52将通过解码编码流的头信息获得的预测模式信息输出到帧内预测部分63、运动补偿部分64和去块滤波器56。 Furthermore, the lossless decoding section 52 is output to the intra prediction section 63 via the prediction mode information decoding coded stream header information obtained, the motion compensation portion 64 and the deblocking filter 56. [0230] 去量化部分53通过对应于图I中的量化部分15的量化系统的系统对通过无损解码部分52解码的量化数据进行去量化。 [0230] dequantizing section 53 corresponds to Figure I by the quantization in the system portion of the system 15 via the lossless decoding section 52 decodes the quantized data dequantized. 反向正交变换部分54通过对应于图I中的正交变换部分14的正交变换系统的系统对去量化部分53的输出进行反向正交变换,并将经过反向正交变换后的数据输出到加法部分55。 Inverse orthogonal transform portion 54 corresponding to FIG. I by the orthogonal transform portion 14 of the orthogonal transform system output system dequantizing portion 53 reverse orthogonal transformation and inverse orthogonal transformation after data is output to the adding section 55. [0231] 加法部分55通过将反向正交变换后的数据与从选择器65提供的预测图像数据加到一起生成解码图像数据,并将解码图像数据输出到去块滤波器56和帧存储器61。 [0231] The addition section 55 by the inverse orthogonal transformed data and the predictive image data supplied from the selector 65 is applied to generate decoded image data together, and outputs the decoded image data to the deblocking filter 56 and frame memory 61 . [0232] 以与图I中的去块滤波器24类似的方式配置去块滤波器56。 [0232] In Fig. I deblocking filter 24 in a similar manner to the deblocking filter 56 configuration. 去块滤波器56通过对从加法部分55提供的解码图像数据进行滤波去除块失真。 Deblocking filter 56 through the decoded image data supplied from the addition section 55 is filtered to remove the block distortion. 然后,去块滤波器56将所述解码图像数据提供到帧存储器61使得在帧存储器61中存储解码图像数据,并将解码图像数据输出到画面重排列缓冲器57。 Then, the deblocking filter 56 to the decoded image data is supplied to the frame memory 61 so that the decoded image data stored in the frame memory 61, and outputs the decoded image data to the picture rearrangement buffer 57. 另外,去块滤波器56基于从无损解码部分52提供的预测模式信息和从下述滤波器设置部分71提供的参数值设置抽头长度和滤波对象像素范围以进行滤波。 In addition, the deblocking filter 56 based on the prediction mode information from the lossless decoding section 52 and the parameter value is set to provide the tap length and the filter object pixel range from the following filter setting section 71, to be filtered. [0233] 画面重排列缓冲器57进行画面重排列。 [0233] screen rearrangement buffer 57. rearrange screen. 具体是,将通过图I的画面重排列缓冲器12重排列的用于编码的帧重排列为初始显示顺序,并将重排列的图像数据输出到D/Α转换部分58。 Specifically, the rearrangement of the frame buffer 12 for encoding rearranged by rearrangement Figure I is the initial screen display order, and outputs the image data rearranged to the D / Α conversion section 58. [0234] D/Α转换部分58对从画面重排列缓冲器57提供的图像数据进行D/Α转换,并通过将图像数据输出到图中未示出的显示器而显示图像。 [0234] D / Α conversion part 58 pairs from the picture rearrangement buffer 57 provides the image data D / Α conversion, and by the output image data to a not shown monitor which displays an image. [0235] 帧存储器61保存从加法部分55提供的滤波前的解码图像数据和从去块滤波器24 提供的滤波后的解码图像数据。 [0235] The frame memory 61 to save the decoded image data from before filtering and adding section 55 provides the decoded image data from the deblocking filter 24 provides a filtered. [0236] 当基于从无损解码部分52提供的预测模式信息对其中已经进行帧内预测的预测块进行解码时,选择器62将从帧存储器61读取的滤波前的解码图像数据提供到帧内预测部分63。 [0236] When based on the prediction mode information from the lossless decoding section 52 has been provided on which intra prediction block prediction decoding, the decoded image data before filtering selector 62 reads from the frame memory 61 are supplied to the frame prediction section 63. 另外,当基于从无损解码部分52提供的预测模式信息对其中已经进行帧间预测的预测块进行解码时,选择器26将从帧存储器61读取的滤波后的解码图像数据提供到运动补偿部分64。 Further, when the prediction mode based on information from the lossless decoding section 52 has been provided on which interframe prediction decoding prediction block selector 26 decoded image data from the frame memory 61 after filtering is supplied to the read motion compensation section 64. [0237] 帧内预测部分63基于从无损解码部分52提供的预测模式信息生成预测图像,并将生成的预测图像数据输出到选择器65。 [0237] Intra prediction information generating portion 63 from the lossless decoding section 52 based on the provided prediction mode prediction image, and generates the prediction image data output to the selector 65. 另外,帧内预测部分63将指示生成的预测图像的块尺寸的信息输出到滤波器设置部分71。 Further, section 63 indicating the intra prediction block size of the generated predicted image information output section 71 to the filter settings. [0238] 运动补偿部分64基于从无损解码部分52提供的预测模式信息进行运动补偿,生成预测图像数据,并将所述预测图像数据输出到选择器65。 [0238] Based on the motion compensation section 64 from the lossless decoding section 52 provides the motion compensated prediction mode information, generating predicted image data and the predictive image data output to the selector 65. 即,基于预测模式信息中包括的运动矢量信息和参考帧信息,运动补偿部分64使用运动矢量,基于运动矢量信息对由参考帧信息指示的参考图像施加运动补偿,并生成预测图像数据。 That is, based on the motion vector information included in the prediction mode information and reference frame information, motion compensation section 64 using a motion vector, based on motion vector information is applied to the motion compensated reference picture indicated by the reference frame information, and generates predictive image data. 另外,运动补偿部分64将指示生成的预测图像的块尺寸的信息输出到滤波器设置部分71。 In addition, the motion compensation section 64 outputs information indicating the block size of the predicted image to generate a filter setting section 71. [0239] 选择器65将在帧内预测部分63中生成的预测图像数据提供到加法部分55。 [0239] Selector 65 to the predictive image data generated at the intra prediction section 63 is supplied to the addition section 55. 另外,选择器65将在运动补偿部分64中生成的预测图像数据提供到加法部分55。 Further, the selector 65 will compensation predicted image data generated in the section 64 is supplied to the addition section 55 movement. [0240] 以与图4中所示的滤波器设置部分41类似的方式配置滤波器设置部分71。 [0240] In the filter shown in FIG. 4 and a similar setting section 41 to set the filter portion 71 is disposed. 另外, 滤波器设置部分71存储指示解码块的预测块尺寸的信息。 Furthermore, filter settings prediction block size storage section 71 indicating the decoded information block. 滤波器设置部分71根据作为解码对象的块和与作为解码对象的块相邻的解码块的相邻边上的预测块尺寸设置抽头长度和滤波对象像素范围。 Filter setting section 71 according to the prediction block size setting and filter tap length range adjacent to the target pixel block to be decoded and the block to be decoded with a decoding block adjacent to the edge. [0241] 滤波器设置部分71将指示设置的抽头长度和设置的滤波对象像素范围的参数值提供到去块滤波器56。 Parameter Value [0241] filter setting section 71 indicates the set of filter tap length and setting target pixel range is supplied to the deblocking filter 56. 另外,当在关注块和相邻块之一的最优模式中的预测块尺寸是扩展块尺寸时,滤波器设置部分71根据更大的预测块尺寸设置抽头长度和滤波对象像素范围。 In addition, when the block of interest and the best mode for one of the adjacent block prediction block size is an extension of the block size, the filter setting section 71 to set the tap length and the filter object pixel range forecast based on a larger block size. 另外,当使用具有大于预定宏块的尺寸的多个宏块时,随着尺寸增大,滤波器设置部分71 设置更长的抽头长度和更大的滤波对象像素范围。 In addition, when a plurality of macro having a size greater than a predetermined macroblock block, as the size increases, the filter setting section 71 to set a longer tap length and larger filter object pixel range. [0242] <6.图像解码装置的操作〉[0243] 下面将参考图12的流程图描述在图像解码装置50中进行的图像解码处理操作。 [0242] <6. The operation of the image decoding apparatus> [0243] below with reference to the flowchart of FIG. 12 described image processing operation in the image decoding apparatus 50. The decoding. [0244] 在步骤ST71,存储缓冲器51存储传输的编码流。 [0244] In step ST71, the storage buffer 51 stores the encoded transport stream. 在步骤ST72,无损解码部分52 进行无损解码处理。 In step ST72, the lossless decoding section 52 performs lossless decoding process. 无损解码单元52对从存储缓冲器51提供的编码流进行解码。 Lossless decoding unit 52 from the storage buffer 51 provides the encoded stream is decoded. S卩,获得通过图I的无损编码部分16编码的每个画面的量化数据。 S Jie, I obtain quantitative data by FIG lossless coding section 16 codes each picture. 另外,无损解码部分52对在编码流的头信息中包括的预测模式信息进行无损解码,并将获得的预测模式信息提供到去块滤波器56和选择器62及65。 Furthermore, the lossless decoding section 52 pairs in the header information encoded stream included lossless decoding prediction mode information, and supplies the prediction mode information obtained to the deblocking filter 56 and selectors 62 and 65. 另外,当预测模式信息是关于帧内预测模式的信息时,无损解码部分52将预测模式信息输出到帧内预测部分63。 Further, when the prediction mode information is the information on the intra prediction mode, the lossless decoding section 52 is output to the intra prediction mode information prediction section 63. 另外,当预测模式信息是关于帧间预测模式的信息时,无损解码部分52将预测模式信息输出到运动补偿部分64。 Further, when the prediction mode information is the information on inter prediction mode, the lossless decoding section 52 outputs prediction mode information to the motion compensation section 64. [0245] 在步骤ST73,去量化部分53进行去量化处理。 [0245] In step ST73, go to the quantization section 53 performs quantization processing. 去量化部分53通过对应于图I中的量化部分15的特征的特征对通过无损解码部分52解码的量化数据进行去量化。 Dequantizing portion 53 corresponding to FIG. I by the quantization characteristic portion 15 is characterized by the lossless decoding section 52 for decoding the quantized data dequantized. [0246] 在步骤ST74,反向正交变换部分54进行反向正交变换处理。 [0246] In step ST74, an inverse orthogonal transformation section 54 inverse orthogonal transform processing. 反向正交变换部分54以对应于图I中正交变换部分14的特征的特征对通过去量化部分53去量化的变换系数数据进行反向正交变换。 Inverse orthogonal transform portion 54 corresponding to the orthogonal transform in Figure I is characterized in the characterizing part 14 through dequantizing quantized transform portion 53 to inverse orthogonal transform coefficient data. [0247] 在步骤ST75,加法部分55生成解码图像数据。 [0247] section 55 generates decoded image data in step ST75, addition. 加法部分55通过将通过进行反向正交变换处理获得的数据与在下述的步骤ST79中选择的预测图像数据加到一起而生成解码图像数据。 Data obtained by the adding section 55 by performing an inverse orthogonal transform processing and added together to generate the following predictive image data selected at step ST79 decoded image data. 从而对初始图像解码。 Thereby decoding the original image. [0248] 在步骤ST76,去块滤波器56进行滤波。 [0248] In step ST76, the deblocking filter 56 for filtering. 去块滤波器56对从加法部分55输出的解码图像数据进行滤波,以去除解码图像中包括的块失真。 Deblocking filter 56 on the decoded image data outputted from the adding section 55 is filtered to remove the block included in the decoded image distortion. [0249] 在步骤ST77,帧存储器61存储解码图像数据。 [0249] In step ST77, the frame memory 61 stores the decoded image data. [0250] 在步骤ST78,帧内预测部分63和运动补偿部分64生成预测图像数据。帧内预测部分63和运动补偿部分64各自生成预测图像数据以对应于从无损解码部分52提供的预测模式信息。 [0251] 具体是,当从无损解码部分52提供用于帧内预测的预测模式信息时,帧内预测部分63基于预测模式信息利用帧存储器61中的解码图像数据进行帧内预测处理,并生成预测图像数据。另外,当从无损解码部分52提供用于帧间预测的预测模式信息时,运动补偿部分64基于预测模式信息利用帧存储器61中的解码图像数据进行运动补偿,并生成预测图像数据。 [0252] 在步骤ST79,选择器65选择预测图像数据。具体是,选择器65选择从帧内预测部分63提供的预测图像和在运动补偿部分64中生成的预测图像数据,并将预测图像数据提供到加法部分55,以在上述步骤ST75中将预测图像数据加到反向正交变换部分54的输出上。 [0253] 在步骤ST80,画面重排列缓冲器57进行画面重排列。具体是,画面重排列缓冲器57将通过图I的图像编码装置10中的画面重排列缓冲器12被重排列以用于编码的帧重排列为原始显示顺序。 [0254] 在步骤ST81,D/Α转换部分58对来自画面重排列缓冲器57的图像数据进行D/A 转换。将该图像输出到图中未示出的显示器并显示该图像。 [0255] 另外,在图12的步骤ST76中的滤波中,滤波器设置部分71进行上述如图10所示的滤波设置处理。滤波器设置部分71根据在作为解码对象的块和与作为解码对象的块相邻的解码块的相邻边上的预测块尺寸设置抽头长度和滤波对象像素范围。当扩展关注块和相邻块中至少一个的相邻边上的块尺寸时,滤波器设置部分71设置扩展的抽头长度和扩展的滤波对象像素范围。另外,当解码相邻块时,在滤波器设置部分71的块尺寸缓冲器中存储在解码块的相邻边上的预测块尺寸。滤波器设置部分71生成指示设置的抽头长度和设置的滤波对象像素范围的参数值,并将参数值提供到去块滤波器56。去块滤波器56以由从滤波器设置部分71提供的参数值指示的抽头长度和滤波对象像素范围对作为解码对象的块与相邻于作为解码对象的块的解码块之间的块边界施加滤波。 [0256] 从而,根据应用本发明的图像解码装置和图像解码方法,在滤波器设置部分71的块尺寸缓冲器中存储在编码处理中使用的指示预测块尺寸的信息。从而,当通过对编码流进行解码生成用于图像显示的解码图像数据时,解码图像中的预测块的位置是清楚的。从而,基于存储在块尺寸缓冲器中的信息,对具有明显块失真的较大块尺寸的部分进行滤波和较高强度的平滑处理,所述滤波被施加到更远离块边界的像素值。从而,可以获得具有不明显块失真的图像质量优异的解码图像。 [0257] 另外,根据解码图像数据是用于生成预测图像的图像数据还是用于图像显示的图像数据,足够使得图像处理装置的滤波器设置部分设置抽头长度和滤波对象像素范围。当图像处理装置例如为图像编码装置时,滤波器设置部分进行对用于生成预测图像的解码图像数据的设置,使得用于生成预测图像的解码图像的图像质量为优异的图像质量,并且编码流的数据量减少。另外,当图像处理装置是图像解码装置时,滤波器设置部分进行设置使得用于图像显示的解码图像的图像质量是用户期望的图像质量。从而,可以当在图像编码装置中提供滤波器设置部分时进行适于图像编码的去块滤波以及进行适于图像解码装置的去块滤波,其中通过滤波器设置部分进行所述去块滤波。 [0258] 可通过硬件、软件或硬件和软件的组合结构执行在本说明书中描述的一系列处理。当通过软件进行处理时,将其中记录处理序列的程序安装到专用硬件中包含的计算机内的存储器中并执行。可选地,可将程序安装到能够进行各种处理的通用计算机中并执行。 [0259] 例如,可以预先将程序记录在作为记录介质的硬盘中或作为记录介质的ROM(只读存储器)中。可选地,可将程序临时地或永久地存储在(记录在)可移动记录介质上,诸如软盘、⑶-ROM (压缩盘只读存储器)、M0 (磁光)盘、DVD (数字通用盘)、磁盘、半导体存储器等。可以将这样的可移动记录介质提供为所谓的封装软件。 [0260] 另外,除了从上述可移动记录介质安装到计算机之外,还可以通过无线电从下载站将程序传输到计算机,或通过诸如LAN (局域网)、互联网等将程序有线传输到计算机。计算机可以接收以这样的方式传输的程序,并将程序安装到诸如内置硬盘等的记录介质上。 [0261] 另外,在上述实施例中,已经描述了这样的情况,其中根据在块边界彼此相邻的相邻块的块尺寸设置抽头长度和滤波对象像素范围作为参数值。然而,当相邻块中的至少一个具有大于预定块尺寸的扩展块尺寸时,滤波器设置部分可以将用于滤波的块边界强度数据的值设置为扩展值,以能够获得具有减少的块失真的优异图像质量的图像。例如,随着相邻块的块尺寸增大,滤波器设置部分将用于滤波的块边界强度数据的值设置为更大,以减少块失真。 [0262] 去块滤波器的处理单位的尺寸或宏块的尺寸不限于本说明书中描述的实例,而是可以是其它尺寸。例如,尽管将H. 264/AVC的宏块的尺寸固定在16X 16像素,HEVC的编码单位的尺寸可以对于每个序列动态指定。 HEVC的编码单位还被称为编码树块。具有最大尺寸的编码单位被称为最大编码单位(LUC)。具有最小尺寸的编码单位被称为最小编码单位(SUC)0通过在设置为图像压缩信息的一部分的序列参数中指定IXU和SCU的尺寸来定义可用编码单位的尺寸范围。另外,通过指定split flag的值识别用于单独序列中的编码单位的尺寸。 [0263] 当编码单位具有正方形时,通过二的幂次(power)表示一边的尺寸。可以将编码单位进一步划分为作为帧内预测和帧间预测的处理单位的预测单位(PU)。另外,可将编码单位划分为作为正交变换的处理单位的变换单位(TU)。除了4X4像素和8X8像素之外, HEVC还允许使用具有16X 16像素和32X32像素的尺寸的变换单位。本说明书中的术语“块”包括宏块、编码单位、预测单位、变换单位或各个其它单位的概念。 [0264] 块的尺寸可以是固定的或者可以动态改变。块尺寸不限于在本说明书描述的实例,而可以为其它尺寸。类似的方法可以用于16X16像素或更小的块,诸如例如4、8和16 的块尺寸。另外,在该情况中,当块尺寸增大时,将滤波器的抽头长度设置为更长或者将滤波对象像素范围设置为更大就足以了。 [0265] 本发明技术不仅可应用于具有4X4像素、8X8像素、16X 16像素以及32X32像素的块尺寸的正方形的情况,还可以应用于8X2像素、2X8像素、16X4像素、4X 16像素、 32X8像素以及8X32像素的非正方形的情况。在该情况中,可以根据相邻块的相邻边的块尺寸设置滤波的抽头长度或滤波对象像素范围。另外,可以根据相邻块的不相邻边的块尺寸设置滤波的抽头长度或滤波对象像素范围。另外,可以根据块的形状和块尺寸适应地选择是应用相邻边上的块尺寸还是应用不相邻边上的块尺寸。 [0266] 另外,用于从编码侧向解码侧传输用于去块滤波器的处理的信息的方法不限于在编码流的头中对这些条信息进行多路复用的方法。例如,可以将这些条信息传输或记录为与编码比特流相关的单独的数据,而不在编码比特流中多路复用。术语“相关”在该情况中表示可以在解码时将比特流中包括的图像(所述图像可以是诸如片段、块等的图像的一部分)与对应于关注图像的信息相互关联。即,可以在与图像(或比特流)不同的传输线上传输信息。另外,可以将信息记录在与图像(或比特流)不同的记录介质(或相同记录介质中的不同记录区域)中。另外,可以诸如多个帧、一个帧或者帧内的一部分的任意单位将信息和图像(或比特流)彼此关联。 [0267] 〈7.应用实例〉[0268] 根据上述实施例的图像编码装置10和图像解码装置50可以应用到:各种电子装置,包括卫星广播中的传输器或接收器、诸如有线电视的有线广播、互联网分发、通过蜂窝通信分发到终端等;用于将图像记录到介质上的记录装置,所述介质诸如为光盘、磁盘、和闪存;或者用于从这些存储介质再现图像的再现装置。下面将描述四个应用实例。 [0269] [7-1.第一应用实例][0270] 图13示出应用上述实施例的电视装置的示意配置的实例。电视装置90包括天线901、调谐器902、解复用器903、解码器904、视频信号处理块905、显示块906、音频信号处理块907、扬声器908、外部接口909、控制块910、用户接口911、以及总线912。 [0271] 调谐器902从经天线901接收的广播信号提取希望信道的信号,并对提取的信号解调。调谐器902然后将通过解调获得的编码比特流输出到解复用器903。 S卩,调谐器902 作为具有用于接收具有编码图像的编码流的电视装置90中的传输装置的功能。 [0272] 解复用器903从编码比特流分离作为观看对象的程序的视频流和音频流,并将每个分离的流输出到解码器904。另外,解复用器903从编码比特流提取EPG(电子节目指南) 的辅助数据等,并将提取的数据提供到控制块910。另外,当编码比特流被加扰时,解复用器903可以进行解扰。 [0273] 解码器904对从解复用器903输入的视频流和音频流进行解码。解码器904然后将通过解码处理生成的视频数据输出到视频信号处理块905。另外,解码器904将通过解码处理生成的音频数据输出到音频信号处理块907。 [0274] 视频信号处理块905再现从解码器904输入的视频数据,并使得在显示块906上显示视频。视频信号处理块905还可以使得在显示块906上显示经网络提供的应用屏。视频信号处理块905还可以根据设置对视频数据进行诸如噪声移除的附加处理。视频信号处理块905还可以生成例如菜单、按钮、光标等的⑶I (图形用户界面)图像,并将生成的图像叠加在输出图像上。 [0275] 通过从视频信号处理块905提供的驱动信号驱动显示块906。显示块906在显示装置(例如液晶显示器、等离子体显示器或0LED)的视频屏幕上显示视频或图像。 [0276] 音频信号处理块907对从解码器904输入的音频数据进行诸如D/Α转换、放大等的再现处理,并使得从扬声器908输出音频。音频信号处理块907还可以对音频数据进行诸如噪声移除的附加处理。外部接口909是用于将电视装置90连接到外部装置或网络的接口。例如,可通过解码器904对经外部接口909接收的视频流或音频流进行解码。 S卩,外部接口909还具有作为用于接收具有编码图像的编码流的电视装置90中的传输装置的功倉泛。 [0277] 控制块910具有诸如CPU (中央处理单元)的处理器和诸如RAM (随机存取存储器) 和ROM (只读存储器)的存储器。存储器存储通过CPU执行的程序、程序数据、EPG数据、经网络获得的数据等。例如在启动电视装置90时通过CPU读取并执行由存储器存储的程序。通过执行程序,CPU根据从例如用户接口911输入的操作信号控制电视装置90的操作。 [0278] 用户接口911被连接到控制块910。用户接口911具有例如按钮和开关以供用户操作电视装置90、用于接收远程控制信号的部分等等。用户接口911检测经这些构成元件的用户的操作,生成操作信号,并将生成的操作信号输出到控制块910。 [0279] 总线912将调谐器902、解复用器903、解码器904、视频信号处理块905、音频信号处理块907、外部接口909以及控制块910相互连接。 [0280] 在这样配置的电视装置90中,解码器904具有根据上述实施例的图像解码装置50 的功能。从而当在电视装置90中解码图像时可以更适当地确定施加去块滤波器的范围,从而可以改善图像质量。 [0281] [7-2.应用的第二实例][0282] 图14示出应用上述实施例的便携电话的示意配置的实例。便携电话920包括天线921、通信块922、音频编译码器923、扬声器924、传声器925、摄像机块926、图像处理块927、多路分解块928、记录和再现块929、显示块930、控制块931、操作块932以及总线933。 [0283] 天线921被连接到通信块922。扬声器924和传声器925被连接到音频编译码器923。操作块932被连接到控制块931。总线将通信块922、音频编译码器923、摄像机块926、图像处理块927、多路分解块928、记录和再现块929、显示块930以及控制块931相互连接。 [0284] 便携电话920在各个操作模式中进行诸如如下的操作:传输和接收音频信号、传输和接收电子邮件或图像数据、拍摄图像、记录数据等,所述操作模式包括语音呼叫模式、 数据通信模式、拍照模式以及视频电话模式。 [0285] 在语音呼叫模式中,将通过传声器925生成的模拟音频信号提供到音频编译码器923。音频编译码器923将模拟音频信号转换为音频数据,对转换的音频数据进行Α/D转换, 并压缩该音频数据。音频编译码器923然后在经过压缩的音频数据输出到通信块922。通信块922对音频数据进行编码和调制以生成传输信号。通信块922然后经天线921将生成的传输信号传输到基站(未示出)。另外,通信块922对经天线921接收的无线电信号进行放大和频率转换以获得接收信号。然后,通信块922通过对接收信号进行解调和解码生成音频数据,并将生成的音频数据输出到音频编译码器923。音频编译码器923对音频数据解压缩并对音频数据进行D/Α转换,以生成模拟音频信号。音频编译码器923然后将生成的音频信号提供到扬声器924以制成音频输出。 [0286] 另外,在数据通信模式中,例如,控制块931根据用户经操作块932的操作生成构成电子邮件的文本数据。控制块931还使得在显示块930上显示文本。另外,控制块931根据经操作块932的来自用户的传输指令生成电子邮件数据,并将生成的电子邮件数据输出到通信块922。通信块922对电子邮件数据进行编码和调制以生成传输信号。通信块922然后经天线921将生成的传输信号传输到基站(未示出)。通信块922还对经天线921接收的无线电信号进行放大和频率转换以获得接收信号。然后,通信块922通过对接收信号进行解调和解码重建电子邮件数据,并将重建的电子邮件数据输出到控制块931。控制块931使得在显示块930上显示电子邮件的内容,并使得将电子邮件数据存储在记录和再现块929 的存储介质上。 [0287] 记录和再现块929具有可任意读取和写入的存储介质。例如,存储介质可以是诸如RAM、闪存等的内置型存储介质,或者可以是诸如硬盘、磁盘、磁光盘、光盘、USB存储器、 存储卡等的外部装载型存储介质。 [0288] 另外,在拍照模式中,例如,摄像机块926通过对主体成像生成图像数据,并将生成的图像数据输出到图像处理块927。图像处理块927对从摄像机块926输入的图像数据编码,并使得在记录和再现块929的存储介质上存储编码流。 [0289] 另外,在视频电话模式中,例如,多路分解块928对通过图像处理块927编码的视频流和从音频编译码器923输入的音频流进行多路复用,并将多路复用流输出到通信块922。通信块922对所述流进行编码和调制以生成传输信号。通信块922然后经天线921 将生成的传输信号传输到基站(未示出)。通信块922还对经天线921接收的无线电信号进行放大和频率转换以获得接收信号。所述传输信号和接收信号可包括编码比特流。然后, 通信块922通过对接收信号进行解调和解码重建数据流,并将重建的数据流输出到多路分解块928。多路分解块928从输入数据流分离视频流和音频流,并将视频流输出到图像处理块927且将音频流输出到音频编译码器923。图像处理块927解码视频流以生成视频数据。视频数据被提供到显示块930。显示块930显示一系列图像。音频编译码器923对音频流解压缩并对音频流进行D/Α转换,以生成模拟音频信号。音频编译码器923然后将生成的音频信号提供到扬声器924以制成音频输出。 [0290] 在这样配置的便携电话920中,图像处理块927具有根据上述实施例的图像编码装置10和图像解码装置50的功能。从而当在便携电话920中编码和解码图像时可以更适当地确定施加去块滤波器的范围,从而可以改善图像质量。 [0291] [7-3.应用的第三实例][0292] 图15示出应用上述实施例的记录和再现装置的示意配置的实例。记录和再现装置940例如对接收的广播程序的音频数据和视频数据编码,并将音频数据和视频数据记录在记录介质上。另外,记录和再现装置940可以例如对从另一装置获得的音频数据和视频数据编码,并将音频数据和视频数据记录在记录介质上。另外,记录和再现装置940例如根据来自用户的指令在监视器和扬声器上再现记录在记录介质上的数据。此时,记录和再现装置940对音频数据和视频数据解码。 [0293] 记录和再现装置940包括调谐器941、外部接口942、编码器943、HDD (硬盘驱动器)944、盘驱动器945、选择器946、解码器947、0SD (屏幕上显示)948、控制块949、以及用户接口950。 [0294] 调谐器941从经天线(未示出)接收的广播信号提取希望信道的信号,并对提取的信号解调。调谐器941然后将通过解调获得的编码比特流输出到选择器946。 S卩,调谐器941具有作为记录和再现装置940中的传输装置的功能。 [0295] 外部接口942是用于将记录和再现装置940连接到外部装置或网络的接口。外部接口942例如可以为IEEE 1394接口、网络接口、USB接口、闪存接口等等。例如,经外部接口942接收的音频数据和视频数据被输入端编码器943。即,外部接口942具有作为记录和再现装置940中的传输装置的功能。 [0296] 当未对视频数据和音频数据编码时,编码器943对从外部接口942输入的视频数据和音频数据编码。编码器943然后将编码比特流输出到选择器946。 [0297] HDD 944在内部硬盘上记录具有压缩在其中的视频、音频等内容数据的编码比特流、各种程序、以及其它数据。 HDD 944还在再现视频和音频时从硬盘读取这些条数据。 [0298] 盘驱动器945在/从装载到其中的记录介质记录/读取数据。装载在盘驱动器945 中的记录介质例如可以为DVD 盘(DVD-视频、DVD-RAM、DVD-R、DVD-RW、DVD+R、DVD+RW 等等)、蓝光光盘(注册商标)等。 [0299] 在记录视频和音频时,选择器946选择从调谐器941或编码器943输入的编码比特流,并将选择的编码比特流输出到HDD 944或盘驱动器945。另外,在再现视频和音频时, 选择器946将从HDD944或盘驱动器945输入的编码比特流输出到解码器947。 [0300] 解码器947对编码比特流解码,并生成音频数据和视频数据。解码器947然后将生成的视频数据输出到OSD 948。另外,解码器904将生成的音频数据输出到外部扬声器。 [0301] OSD 948再现从解码器947输入的视频数据,并显示视频。 OSD 948还在显示视频上重叠例如菜单、按钮、光标等的⑶I图像。 [0302] 控制块949具有诸如CPU的处理器和诸如RAM和ROM的存储器。存储器存储通过(PU执行的程序、程序数据等。例如在启动记录和再现装置940时通过CPU读取并执行由存储器存储的程序。通过执行程序,CPU根据从例如用户接口950输入的操作信号控制记录和再现装置940的操作。[0303] 用户接口950被连接到控制块949。用户接口950具有例如按钮和开关以供用户操作记录和再现装置940、用于接收远程控制信号的部分等等。用户接口950检测经这些构成元件的用户的操作,生成操作信号,并将生成的操作信号输出到控制块949。[0304] 在这样配置的记录和再现装置940中,编码器943具有根据上述实施例的图像编码装置10的功能。另外,解码器947具有根据上述实施例的图像解码装置50的功能。从而当在记录和再现装置940中编码和解码图像时可以更适当地确定施加去块滤波器的范围, 从而可以改善图像质量。[0305] [7-4.应用的第四实例][0306] 图16示出应用上述实施例的成像装置的示意配置的实例。成像装置960通过对主体成像生成图像,对图像数据编码,并将图像数据记录在记录介质上。[0307] 成像装置960包括光学块961、成像块962、信号处理块963、图像处理块964、显示块965、外部接口966、存储器967、介质驱动器968、OSD 969、控制块970、用户接口971、以及总线972。[0308] 光学块961被连接到成像块962。成像块962被连接到信号处理块963。显示块965被连接到图像处理块964。用户接口971被连接到控制块970。总线972将图像处理块964、外部接口966、存储器967、介质驱动器968、OSD 969以及控制块970相互连接。[0309] 光学块961具有聚焦透镜、光圈机构等。光学块961在成像块962的成像表面上形成主体的光学图像。成像块962具有CXD或CMOS图像传感器等。成像块962通过光电转换将在成像表面上形成的光学图像转换成作为电信号的图像信号。成像块962然后将图像信号输出到信号处理块963。[0310] 信号处理块963对从成像块962输入的图像信号进行各种摄像机信号处理,诸如拐点校正(knee correction)、伽马校正、颜色校正等。信号处理块963将经过摄像机信号处理的图像数据输出到图像处理块964。[0311] 图像处理块964对从信号处理块963输入的图像数据进行编码以生成编码数据。图像处理块964然后将生成的编码数据输出到外部接口966或介质驱动器968。图像处理块964还对从外部接口966或介质驱动器968输入的编码数据进行解码,以生成图像数据。图像处理块964然后将生成的图像数据输出显示块965。图像处理块964还可以将从信号处理块963输入的图像数据输出到显示块965以显示图像。图像处理块964还可以在输出到显示块965的图像上重叠从OSD 969获得的用于显示的数据。 [0312] OSD 969生成例如菜单、按钮、光标等的⑶I图像,并将生成的图像输出到图像处理块964。 [0313] 外部接口966被配置为例如输入-输出终端。外部接口966在例如打印图像时将成像装置960连接到打印机。另外,根据需要将外部接口966与驱动器连接。例如将诸如磁盘或光盘的可移动介质装载到驱动器。可将从可移动介质读取的程序安装到成像装置960上。另外,外部接口966可被配置网络接口,其将被连接到诸如LAN、互联网等等的网络。即,外部接口966具有作为成像装置960中的传输装置的功能。 [0314] 被装载到介质驱动器968中的记录介质可以是可任意读取和写入的可移动介质, 诸如磁盘、磁光盘、光盘、或半导体存储器。另外,可以以固定方式将记录介质安装到介质驱动器968中,以形成非便携的存储部分,诸如内置型硬盘驱动器或SSD (固态驱动器)。 [0315] 控制块970具有诸如CPU的处理器和诸如RAM和ROM的存储器。存储器存储通过CN 102934429 A书明说24/25 页(PU执行的程序、程序数据等。例如在启动成像装置960时通过CPU读取并执行由存储器存储的程序。通过执行程序,CPU根据从例如用户接口971输入的操作信号控制成像装置960 的操作。[0316] 用户接口971被连接到控制块970。用户接口971具有例如按钮和开关以供用户操作成像装置960等。用户接口971检测经这些构成元件的用户的操作,生成操作信号,并将生成的操作信号输出到控制块970。[0317] 在这样配置的成像装置960中,图像处理块964具有根据上述实施例的图像编码装置10和图像解码装置50的功能。从而当在成像装置960中编码和解码图像时可以更适当地确定施加去块滤波器的范围,从而可以改善图像质量。[0318] 另外,本发明不应被解释为限于上述实施例。所述实施例以示意的形式公开了本发明技术。显然,在不偏离本发明的主旨的情况下,本领域技术人员可以对所述实施例进行修改和替换。即,考虑权利要求,以确定本发明的主旨[0319] 另外,本发明还可以采用下面的组成。[0320] (1) 一种图像处理装置,包括:[0321] 解码部分,用于对在每个块中的编码的图像数据进行解码;[0322] 滤波器,用于对通过所述解码部分解码的解码图像数据施加用于去除块失真的滤波;以及[0323] 滤波器设置部分,用于根据在块边界处相邻的相邻块的块尺寸设置用于块边界的滤波的抽头长度或者作为滤波对象的滤波对象像素范围。[0324] (2)根据(I)所述的图像处理装置,其中当将相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将所述抽头长度设置为扩展的长度。[0325] (3)根据(2)所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将滤波器的抽头长度设置得更长。[0326] (4)根据(I)到(3)中任一项所述的图像处理装置,其中当相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将所述滤波对象像素范围设置为扩展的宽度。[0327] (5)根据(I)到(4)中任一项所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将所述滤波对象像素范围设置得更宽。[0328] (6)根据(I)到(5)中任一项所述的图像处理装置,其中当相邻块中的至少一个被扩展为大于预定块尺寸的尺寸时,所述滤波器设置部分将用于滤波的块边界强度数据的值设置为扩展的值。 [0329] (7)根据(6)所述的图像处理装置,其中随着所述相邻块的块尺寸增大,所述滤波器设置部分将用于滤波的块边界强度数据的值设置得更大。 [0330] (8)根据(I)到(7)中任一项所述的图像处理装置,其中所述滤波器设置部分根据相邻块的相邻边上的块尺寸来设置所述滤波的抽头长度或滤波对象像素范围。 [0331] (9)根据(I)到(8)中任一项所述的图像处理装置,其中所述滤波器设置部分根据与相邻块的块尺寸对应的情况分类来设置所述滤波的抽头长度和滤波对象像素范围。 [0332] (10)根据(9)的图像处理装置,其中所述情况分类为所述相邻块都具有预定块尺寸或更小的块尺寸的情况、以及所述相邻块中的至少一个被扩展为大于预定块尺寸的尺寸27的情况。 [0333] (11)根据(10)所述的图像处理装置,其中所述滤波器设置部分进行情况分类,以分类为相邻块为16X16像素或更小的情况、两个块中的至少一个大于16X16像素且两个块都为32X32像素或更小的情况、以及两个块中的至少一个大于32X32像素的情况。 [0334] (12)根据(I)到(11)中任一项所述的图像处理装置,其中当进行帧内预测或帧间预测时,所述块尺寸是作为处理单位的预测块尺寸。 [0335] (13)根据(I)到(12)中任一项所述的图像处理装置,其中当进行正交变换时,所述块尺寸是作为处理单位的变换尺寸。 [0336] (14)根据(2)到(13)中任一项所述的图像处理装置,其中所述预定块尺寸是H. 264/AVC标准的宏块尺寸。 [0337] (15)根据(I)到(14)中任一项所述的图像处理装置,其中所述滤波器设置部分根据解码图像数据是用于生成预测图像的图像数据还是用于图像显示的图像数据来设置抽头长度或滤波对象像素范围。 [0338] 工业适用性[0339] 根据本发明的图像处理装置和图像处理方法可以提供具有减少的块失真的图像质量优异的图像。本发明技术从而适用于当经诸如卫星广播、有线电视、互联网、便携电话等的网络介质传输和接收通过在MPEG、H. 26x等的块单元中进行编码获得的图像信息(比特流)时、或者当在诸如光盘、磁盘、闪存等的存储介质上处理图像信息时使用的图像编码装置、图像解码装置等。 [0340] 附图标记的描述[0341] 10 图像编码装置,11 Α/D转换部分,12、57 画面重排列缓冲器,13 减法部分,14 正交变换部分,15 量化部分,16 无损编码部分,17、51 存储缓冲器, 18 速率控制部分,21、53 去量化部分,22、54 反向正交变换部分,23、55加法部分,24、56 去块滤波器,25、61 帧存储器,26、62、65 选择器,31、63 帧内预测部分,32 运动预测和补偿部分,33 预测图像和最优模式选择部分,41、71 滤波器设置部分,50 图像解码装置,52 无损解码部分,58 D/Α转换部分,64 运动补偿部分,90 电视装置,92 便携电话,94 记录和再现装置,96 成像装置,241 滤波强度确定部分,242 滤波部分,412 参数值生成部分,901、921 天线,902、941调谐器,903 解复用器,904、947 解码器,905 视频信号处理快,906 显示块,907 音频信号处理块,908 扬声器,909、942、966 外部接口块,912、933、972 总线,922 通信块,923 音频编译码器,924 扬声器,925 传声器,926摄像机块,927 图像处理块,928 多路分解块,929 记录和再现块,930 显示块,943 编码器,944 HDD块,945 盘驱动器,948、969 OSD块,961 光学块,962 成像块,963摄像机信号处理块,964 图像数据处理块,965 显示块,967存储块,968 介质驱动器

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
CN1607836A *14 Oct 200420 Apr 2005松下电器产业株式会社De-blocking filter processing apparatus and de-blocking filter processing method
CN1820512A *10 Mar 200516 Aug 2006学校法人大洋学园Method, medium, and filter removing a blocking effect
CN101502119A *31 Jul 20075 Aug 2009汤姆逊许可公司Adaptive geometric partitioning for video decoding
TW200913724A * Title not available
US20090129478 *14 Nov 200821 May 2009Stmicroelectronics SaDeblocking filter
US20100080472 *22 Sep 20091 Apr 2010Kabushiki Kaisha ToshibaImage processing apparatus, moving image decoding apparatus, moving image encoding apparatus and method
WO2010001911A1 *30 Jun 20097 Jan 2010シャープ株式会社Filter device
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
CN103220529A *15 Apr 201324 Jul 2013北京大学Method for achieving video coding and decoding loop filtering
CN103220529B *15 Apr 201324 Feb 2016北京大学一种视频编解码环路滤波的实现方法
CN103246992A *30 May 201314 Aug 2013叶旭东3D (three-dimensional) house display system, on-line house display system and on-line house sales method
WO2017054194A1 *30 Sep 20156 Apr 2017Realnetworks, Inc.Layered deblocking filtering in video processing systems and methods
Classifications
International ClassificationH04N7/26
Cooperative ClassificationH04N19/82, H04N19/17, H04N19/117, H04N19/60, H04N19/44, H04N19/86, H04N19/176, H04N19/80, H04N19/157
Legal Events
DateCodeEventDescription
13 Feb 2013C06Publication
20 Mar 2013C10Entry into substantive examination
28 Sep 2016C14Grant of patent or utility model