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 numberCN1685357 A
Publication typeApplication
Application numberCN 03822593
PCT numberPCT/CN2003/000816
Publication date19 Oct 2005
Filing date25 Sep 2003
Priority date25 Sep 2002
Also published asCN100380391C, WO2004029862A1
Publication number03822593.X, CN 03822593, CN 1685357 A, CN 1685357A, CN-A-1685357, CN03822593, CN03822593.X, CN1685357 A, CN1685357A, PCT/2003/816, PCT/CN/2003/000816, PCT/CN/2003/00816, PCT/CN/3/000816, PCT/CN/3/00816, PCT/CN2003/000816, PCT/CN2003/00816, PCT/CN2003000816, PCT/CN200300816, PCT/CN3/000816, PCT/CN3/00816, PCT/CN3000816, PCT/CN300816
Inventors张大鹏戴维, 钮旋, 卢光明, 江伟健亚当, 王明强
Applicant香港理工大学
Export CitationBiBTeX, EndNote, RefMan
External Links: SIPO, Espacenet
Method and apparatus for identifying a palmprint image
CN 1685357 A
Abstract  translated from Chinese
一种掌纹识别方法,包括:分析来自手掌图像的区域,以利用该区域获得皮肤表面的纹理数据。 A palmprint recognition method, comprising: analysis from the palm area of the image in order to take advantage of the region to get the skin surface texture data. 将纹理数据与数据库中的参考信息进行比较,以确定对个体的识别。 The texture data in the database is compared with reference information to determine identification of the individual. 一种用于捕获手掌图像的设备,包括:外壳,其中具有窗口;以及设置在所述外壳中的图像传感器和光源,用于通过窗口来捕获图像。 A method for capturing an image of the palm device, comprising: a housing having a window; and an image sensor and a light source disposed in said housing for capturing images through the window. 在表面上设置凸起。 On the surface of the projection. 设置所述凸起,以使其位于适当放置在窗口上的手的已知相邻位置,以便捕获包括手的手掌区域的图像。 Setting said projections, so that it is appropriately placed on the window the known hand adjacent positions, in order to capture an image including the hand palm area.
Claims(19)  translated from Chinese
1.一种生物统计学识别方法,包括:从个体获得皮肤表面区域的图像;分析所述图像,以提取出皮肤表面区域上的纹理特征;以及将所述纹理特征与数据库中的参考信息进行比较。 A biometric identification method, comprising: obtaining individual images from the skin surface area; analyzing the images to extract the texture on the skin surface area; and the texture and reference information in the database Compare.
2.一种生物统计学识别方,包括:获得个体的手的部分内表面的图像;获得手的内表面的指定区域内的皮肤表面的子图像;分析所述子图像,以获得皮肤表面的纹理数据;以及将所述纹理数据与数据库中的参考信息进行比较。 2. A biometric identifying party, comprising: obtaining an image of the inner surface of the individual's hand; sub-images obtained hand skin surface within the designated area of the inner surface; analyzing said sub-image, to obtain a skin surface texture data; and the texture data and reference information in the database for comparison.
3.根据权利要求2所述的方法,其特征在于所述指定区域依赖于手的一个或多个特征。 3. The method according to claim 2, characterized in that the hand region is dependent on one or more of the specified feature.
4.根据权利要求2或3所述的方法,其特征在于所述一个或多个特征是手指之间区域。 4. The method according to claim 2 or claim 3, wherein, characterized in that said one or more features are regions between the fingers.
5.根据前述权利要求之一所述的方法,其特征在于通过以下步骤获得所述子图像,包括:识别表示手指之间区域的至少两个点;确定具有第一和第二轴的坐标系统,其中所述两个点位于所述第一轴上,且与所述第二轴等距;以及利用所述两个点之间的距离,确定所述子图像在所述坐标系统内的参数。 5. A process according to claim one of the preceding claims, characterized in that the sub-images obtained by the steps comprising: identifying at least two points represents the area between the fingers; determining a coordinate system having a first and second shafts , wherein the two points are located on said first shaft and said second shaft and equidistant; and using the distance between the two points, determining the parameters within the sub-image coordinate system .
6.根据权利要求5所述的方法,其特征在于所述子图像的所述参数包括所述坐标系统中、以(0.25D,0.5D)、(1.25D,0.5D)、(0.25D,-0.5D)和(1.25D,-0.5D)表示的点,其中D是所述两个点之间的距离。 6. The method of claim 5, wherein said parameter comprises the sub-image coordinate system, to (0.25D, 0.5D), (1.25D, 0.5D), (0.25D, -0.5D) point and the (1.25D, -0.5D) represented, where D is the distance between the two points.
7.根据权利要求5或6所述的方法,其特征在于还包括对所述子图像进行规一化的步骤。 7. A method according to claim 5 or 6, wherein, characterized by further comprising the sub-image normalization step.
8.根据前述权利要求之一所述的方法,其特征在于分析所述子图像包括利用伽柏滤波器。 8. Method according to one of the preceding claims, characterized in that the analysis of the sub-image including the use of Gabor filters.
9.根据前述权利要求之一所述的方法,其特征在于分析所述子图像包括利用伽柏分析以低分辨率分割所述子图像的层。 9. Method according to one of the preceding claims, characterized in that the analysis comprises the use of said sub-image segmentation of the Gabor analysis at a low resolution sub-image layer.
11.根据前述权利要求之一所述的方法,其特征在于将所述子图像分割为两个部分,实部和虚部,将每一部分存储为向量。 11. The method according to one of the preceding claims, characterized in that said sub-image is divided into two parts, the real and imaginary parts, each part will be stored as a vector.
12.根据权利要求11所述的方法,其特征在于将所述纹理数据与数据库中的参考信息进行比较基于以下形式的汉明距离:D0=Σi=1NΣj=1NPM(i,j)∩QM(i,j)((PR(i,j)⊗QR(i,j)+PI(i,j)⊗QI(i,j)))2Σi=1NΣj=1NPM(i,j)∩QM(i,j),]]>其中PR(QR)和PI(QI)是所述实部和所述虚部。 12. The method according to claim 11, characterized in that the texture data in the database is compared with reference information on the form of the Hamming distance: D0 = & Sigma; i = 1N & Sigma; j = 1NPM (i, j ) & cap; QM (i, j) ((PR (i, j) & CircleTimes; QR (i, j) + PI (i, j) & CircleTimes; QI (i, j))) 2 & Sigma; i = 1N & Sigma; j = 1NPM (i, j) & cap; QM (i, j),]]> where PR (QR) and PI (QI) is the real part and the imaginary part.
13.一种掌纹图像捕获设备,包括:外壳,其中具有窗口;设置在所述外壳中的图像传感器,用于通过窗口来捕获图像;光源,用于照亮所述窗口;以及与所述窗口相邻的至少一个凸起,其中设置所述凸起,以使其位于适当放置在窗口上的手的已知相邻位置,以便捕获包括手的手掌区域的图像。 13. A palmprint image capture device, comprising: a housing having a window; disposed in the housing of the image sensor for capturing an image through the window; a light source for illuminating said windows; as well as the at least one projection adjacent to the window, wherein the projections provided, so that it is appropriately placed on the window the known hand adjacent positions, in order to capture an image including the hand palm area.
14.根据权利要求13所述的设备,其特征在于所述凸起是设置在适当放置在窗口上的手的两个或多个手指之间的栓或销。 14. The apparatus according to claim 13, wherein said projection is disposed between two or more appropriately placed on the finger of the hand of the window bolt or pin.
15.根据权利要求13或14所述的方法,其特征在于所述光源是所述图像传感器位于其中心的环面。 13 or 15. The method according to claim 14, wherein said light source is located in the center of the image sensor of the torus.
16.根据权利要求13到15之一所述的方法,其特征在于所述图像传感器是电荷耦合器件或互补金属氧化物半导体传感器。 16. The method according to one of claim 15, wherein said image sensor is a charge coupled device or a complementary metal oxide semiconductor sensor.
17.一种掌纹图像捕获设备,包括:外壳,其中具有窗口;设置在所述外壳中的图像传感器,用于通过窗口来捕获图像;光源,用于照亮所述窗口;控制器,用于控制所述图像传感器和光源,以便捕获图像;以及与所述窗口相邻的至少一个凸起,其中设置所述凸起,以使其位于适当放置在窗口上的手的已知相邻位置,以便捕获包括手的手掌区域的图像。 17. A palmprint image capture device, comprising: a housing having a window; disposed in the housing of the image sensor for capturing an image through the window; a light source for illuminating said window; controller, with controlling said image sensor and a light source, in order to capture an image; and a window adjacent to said at least one projection, wherein the projection set, so that it is appropriately placed on the window the known hand adjacent positions , in order to capture an image of the palm region including hands.
18.根据权利要求17所述的设备,其特征在于所述凸起是设置在适当放置在窗口上的手的两个或多个手指之间的栓或销。 18. The apparatus according to claim 17, wherein said projection is disposed between two appropriately placed on the window or more fingers of the hand bolt or pin.
19.根据权利要求17或18所述的方法,其特征在于所述光源是所述图像传感器位于其中心的环面。 17 or 19. The method according to claim 18, wherein said light source is located in the center of the image sensor of the torus.
20.根据权利要求17到19之一所述的方法,其特征在于所述图像传感器是电荷耦合器件或互补金属氧化物半导体传感器。 20. The method according to one of claim 19, wherein said image sensor is a charge coupled device or a complementary metal oxide semiconductor sensor.
Description  translated from Chinese
掌纹识别方法和设备 Palmprint recognition method and apparatus

技术领域 FIELD

本发明涉及生物统计学识别,更具体地涉及一种分析掌纹以便识别个体的方法。 The present invention relates to biometric identification, and more particularly to a palmprint analysis method to identify an individual. 本发明还涉及用于捕获掌纹图像以便识别个体的设备。 The present invention also relates to a palmprint image capturing device to identify the individual.

背景技术 BACKGROUND

利用掌纹识别作为一种个人识别方法是一种代替指纹的新生物统计学技术。 Use of palmprint recognition as a personal identification method is an alternative to a new biometric fingerprint technology. 已知的方法包括分析掌纹,以识别掌纹图像中的奇点、细节和皱纹。 Known methods include palmprint analysis to identify the palmprint image singularities, details and wrinkles. 这些已知的方法需要如图1所示的高分辨率图像。 These known methods require a high resolution image as shown in Fig. 这可以通过染色的掌纹来获得。 This can be obtained by staining palm. 但是,这样做比较肮脏,并且不能获得实时识别。 However, this relatively dirty, and can not get real-time identification.

为了克服染色掌纹的问题,一些公司已经开发了高分辨率掌纹扫描仪和识别系统。 To overcome the problem of stained palm, some companies have developed a high-resolution palmprint scanners and identification system. 但是,这些用于捕获高分辨率图像的设备是昂贵的,且依赖于高性能的计算机来满足实时识别的需要。 However, these devices for capturing a high resolution image is expensive, and depends on high performance computers to meet the needs of real-time identification.

对于上述问题的一种解决方法在于使用低分辨率图像。 One solution to these problems lies in the use of low-resolution images. 图2示出了与图1相对应的低分辨率图像。 Figure 2 shows a diagram corresponding to a low-resolution image. 但是,在低分辨率图像中,不能够容易地观察到奇点和细节,因而,更容易识别的皱纹必须在识别中发挥重要作用。 However, in low-resolution images, can not be easily observed singularities and detail, thus, easier to identify the wrinkles must play an important role in the recognition. 但是,从图2中可以注意到,只有小部分周围较为清楚,问题在于,其是否提供了足够的独特性,以便在大量人口中,可靠地识别个体。 However, it can be noted from Figure 2, only a relatively small portion around the clear, the problem is that, if it offers a sufficient uniqueness, so that a large number of the population, to reliably identify the individual.

发明内容 SUMMARY

本发明的一个目的在于提供一种生物统计学识别方法,更具体地,一种分析掌纹以便识别个体的方法,其克服或改善了现有方法的缺陷。 An object of the present invention is to provide a method of biometric identification, and more particularly, a palmprint analysis method to identify an individual, which overcome or ameliorate the disadvantages of the prior methods. 本发明的另一目的是提供一种用于捕获掌纹图像的设备,其克服或改善了现有设备的缺陷,或者至少其向公众提供了一种有用的可选设备。 Another object of the present invention is to provide a device for capturing palmprint image, which overcome or ameliorate the shortcomings of existing equipment, or at least provides a useful alternative to the public device.

根据本发明的第一方面,提出了一种生物统计学识别方法,包括:从个体获得皮肤表面区域的图像;分析所述图像,以提取出皮肤表面区域上的纹理特征;以及将所述纹理特征与数据库中的参考信息进行比较。 According to a first aspect of the present invention, there is proposed a method of biometric identification, comprising: obtaining an image from the skin surface area of an individual; analyzing the image, to extract texture features on the skin surface area; and the texture Features and reference information in the database for comparison.

根据本发明的第二方面,提出了一种生物统计学识别方法,包括:获得个体的手的部分内表面的图像;获得手的内表面的指定区域内的皮肤表面的子图像;分析所述子图像,以获得皮肤表面的纹理数据;以及将所述纹理数据与数据库中的参考信息进行比较。 According to a second aspect of the present invention, there is proposed a method of biometric identification, comprising: obtaining an image of an individual's hand portion of the inner surface; sub-image of the skin surface to obtain an inner surface of the hand within a designated area; Analysis of the sub-image, to obtain texture data of the skin surface; and the texture data in the database is compared with reference information.

优选地,所述指定区域依赖于手的一个或多个特征。 Preferably, the region is dependent on the hand of the one or more features specified.

优选地,所述一个或多个特征是手指之间区域。 Advantageously, said one or more features are regions between the fingers.

优选地,通过以下步骤获得所述子图像,包括:识别表示手指之间的区域的至少两个点;确定具有第一和第二轴的坐标系统,其中所述两个点位于所述第一轴上,且与所述第二轴等距;以及利用所述两个点之间的距离,确定所述子图像在所述坐标系统内的参数。 Preferably, is obtained by said sub-image, comprising: identifying at least two points represents the area between the fingers; determining a coordinate system having a first and second shafts, wherein the two points located on the first shaft, and equidistant from said second axis; and using the distance between the two points, determining the parameters within the sub-image coordinate system.

优选地,所述子图像的所述参数包括所述坐标系统中、以(0.25D,0.5D)、(1.25D,0.5D)、(0.25D,-0.5D)和(1.25D,-0.5D)表示的点,其中D是所述两个点之间的距离。 Preferably, said parameter comprises the sub-image coordinate system, to (0.25D, 0.5D), (1.25D, 0.5D), (0.25D, -0.5D) and (1.25D, -0.5 D) represented by the point, where D is the distance between the two points.

优选地,还包括对所述子图像进行规一化的步骤。 Preferably, further comprising the sub-image normalization step.

优选地,分析所述子图像包括利用伽柏滤波器。 Preferably, analysis of the sub-image including the use of Gabor filters.

优选地,分析所述子图像包括利用伽柏分析以低分辨率分割所述子图像的层。 Preferably, analysis of the sub-image including the use of a Gabor analysis at a low resolution of the sub-images divided layers.

优选地,将所述子图像分割为两个部分,实部和虚部,将每一部分存储为向量。 Preferably, the sub-image is divided into two parts, the real and imaginary parts, each part will be stored as a vector.

优选地,将所述纹理数据与数据库中的参考信息进行比较基于以下形式的汉明距离: Preferably, the texture data in the database were compared based on the following reference information in the form of a Hamming distance:

D0=Σi=1NΣj=1NPM(i,j)∩QM(i,j)((PR(i,j)⊗QR(i,j)+PI(i,j)⊗QI(i,j)))2Σi=1NΣj=1NPM(i,j)∩QM(i,j),]]>其中PR(QR)和PI(QI)和是所述实部和所述虚部。 D0 = & Sigma; i = 1N & Sigma; j = 1NPM (i, j) & cap; QM (i, j) ((PR (i, j) & CircleTimes; QR (i, j) + PI (i, j) & CircleTimes; QI (i, j))) 2 & Sigma; i = 1N & Sigma; j = 1NPM (i, j) & cap; QM (i, j),]]> where PR (QR) and PI (QI), and are the real and The imaginary part.

根据本发明的第三方面,提出了一种掌纹图像捕获设备,包括:外壳,其中具有窗口;设置在所述外壳中的图像传感器,并通过窗口来捕获图像;光源,用于照亮所述窗口;以及与所述窗口相邻的至少一个凸起,其中设置所述凸起,以使其位于适当放置在窗口上的手的已知相邻位置,以便捕获包括手的手掌区域的图像。 According to a third aspect of the present invention, there is proposed a palmprint image capturing apparatus, comprising: a housing having a window; disposed in the housing of the image sensor, and captures an image through the window; a light source for illuminating the said window; and a window adjacent to the at least one projection, wherein the projection disposed, so that it is appropriately placed on the window the known hand adjacent positions, in order to capture an image including the hand palm area .

根据本发明的第四方面,提出了一种掌纹图像捕获设备,包括:外壳,其中具有窗口;设置在所述外壳中的图像传感器,并通过窗口来捕获图像;光源,用于照亮所述窗口;控制器,用于控制所述图像传感器和光源,以便捕获图像;以及与所述窗口相邻的至少一个凸起,其中设置所述凸起,以使其位于适当放置在窗口上的手的已知相邻位置,以便捕获包括手的手掌区域的图像。 According to a fourth aspect of the present invention, there is proposed a palmprint image capturing apparatus, comprising: a housing having a window; disposed in the housing of the image sensor, and captures an image through the window; a light source for illuminating the said window; a controller for controlling the image sensor and a light source, in order to capture an image; and a window adjacent to said at least one projection, wherein the projection set, so that it is appropriately placed on the window Hand known adjacent positions, in order to capture an image including the hand palm area.

优选地,所述凸起是设置在适当放置在窗口上的手的两个或多个手指之间的栓或销。 Preferably, the protrusion is disposed between two appropriately placed on a window or a plurality of fingers of the hand of the bolt or pin.

优选地,所述光源是所述图像传感器位于其中心的环面。 Preferably, the light source is located in the center of the image sensor of the torus.

优选地,所述图像传感器是电荷耦合器件(CCD)或互补金属氧化物半导体(CMOS)传感器。 Preferably, the image sensor is a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) sensor.

通过仅作为示例给出的以下描述,本发明的其他方面将变得显而易见。 From the following description given by way of example only, other aspects of the invention will become apparent.

附图说明 Brief Description

现在,将参照附图,对本发明的实施例进行描述,其中: Now, with reference to the drawings, embodiments of the present invention will be described, wherein:

图1示出了典型的高分辨率掌纹图像;图2示出了典型的低分辨率掌纹图像;图3到图8示出了对手内侧图像的预处理;图9和图10示出了手在手掌读取器上的不正确放置及相应的预处理图像;图11到图14示出了预处理图像、实部和虚部以及掩模;图15和图16示出了第一和第二收集图像之间图像质量的差别;图17和图18示出了根据本发明的方法的验证测试结果;图19示出了根据本发明的掌纹图像捕获设备的示意图;图20示出了该设备的图像捕获表面的平面图;图21是沿图19中的A-A'得到的剖面图,其中CCD摄像机沿圆形光旋转;以及图22示出了由设备所捕获的原始手掌图像。 Figure 1 shows a typical high-resolution palmprint image; FIG. 2 shows a typical low-resolution palmprint image; FIG. 3 to FIG. 8 shows an image of the inside of the opponent preprocessing; FIG. 9 and FIG. 10 shows incorrectly placed his hand on the palm reader and the corresponding image preprocessing; Figs. 11 to 14 shows a pre-processing the image, and the real and imaginary parts of the mask; Figure 15 and Figure 16 shows the first and the difference between the second image quality captured images; Fig. 17 and FIG. 18 shows a method of verification test result of the present invention; FIG. 19 shows a schematic view of the palm image capturing apparatus according to the present invention; FIG. 20 shows a plan view of the image capturing surface of the apparatus; Fig. 21 Fig. 19 along the A-A 'sectional view obtained, wherein the CCD camera along a circular light rotation; and Figure 22 shows the device captured the original palm images.

具体实施方式 DETAILED DESCRIPTION

本发明的掌纹识别方法包括三个部分:1)获得个体的掌纹图像;2)根据该图像,分析皮肤纹理数据;以及3)将皮肤纹理数据与存储在数据库中的信息进行比较。 Palmprint identification method of the invention comprises three parts: 1) to obtain individual palmprint image; 2) According to the image, analyzing the skin texture data; and 3) the information of the skin texture data stored in the database for comparison. 下面,将更为详细的描述这些步骤。 Hereinafter, a more detailed description of these steps.

1)获得个体的掌纹图像参考图3,利用CCD摄像机,按照已知的方式来获得手的部分内表面的低分辨率图像。 1) obtaining individual palmprint image with reference to Figure 3, a CCD camera, in a known manner to obtain the low-resolution image of the inner surface of the hand. 为了从图像中提取出识别数据,必须利用手的特征来识别手掌区域的可重复子图像。 In order to extract the identification data from the image, the hand must be used to identify the characteristics of the palm region may be repeated sub-image. 在优选实施例中,识别手指间的缺口,并用作构建坐标系统的参数,可以在所述坐标系统中找出定义了子图像的参数。 Embodiment, the gap between the finger or in a preferred embodiment, and the parameters used to build the coordinate system, can find out the parameters defined in the sub-image coordinate system. 优选实施例具有六个主要步骤,如以下所述。 Preferred embodiment has six major steps, as described below.

参照图4,第一步是对原始图像0(x,y)应用低通滤波器L(u,v),如高斯滤波器等。 Referring to Figure 4, the first step is the original image 0 (x, y) low-pass filter L (u, v), such as a Gaussian filter or the like. 然后,利用阈值Tp,将卷积的图像转换为二值化图像B(x,y)。 Then, using the threshold Tp, to convert the image convolution binarized image B (x, y).

参照图5,第二步是利用边界跟踪算法,获得手指之间的缺口的边界(Fixj,Fiyj):其中i=1、2。 Referring to FIG. 5, the second step is the use of boundary tracking algorithm, to obtain the gap between the boundaries of the finger (Fixj, Fiyj): where i = 1,2. 并不提取出无名指与中指之间的缺口的边界,由于其对于以下处理没有用。 Not extracted boundary gap between the ring finger and middle finger, not because of its use for the following process.

参照图6,第三步是计算缺口(Fixj,Fiyj)的切线。 Referring to FIG. 6, the third step is to calculate the gap (Fixj, Fiyj) tangent. 如果(x1,y1)和(x2,y2)分别是(F1xj,F1yj)和(F2xj,F2yj)上的两个点,则对于所有的i和j,通过这两个点的直线(y=mx+c)满足不等式Fiyj≤mFixj+C。 If (x1, y1) and (x2, y2) are (F1xj, F1yj) and (F2xj, F2yj) on two points, then for all i and j, the two points by a straight line (y = mx + c) satisfies the inequality Fiyj≤mFixj + C. 直线(y=mx+c)是两个缺口的切线。 A straight line (y = mx + c) is tangent to two notches. 以图6中的数字2表示的这条直线是坐标系统的Y轴,用于确定子图像1的位置。 In Figure 6, the number 2 of this straight line is the Y-axis of the coordinate system for determining the position of a sub-image of.

第四步是找出通过两个点的中点、垂直于直线2的直线3,以确定坐标系统的X轴和原点。 The fourth step is to find the two points by a middle point, straight line perpendicular to the straight line 2, 3, in order to determine the X-axis and the origin of the coordinate system. 所述两个点位于Y轴上,且与X轴等距。 The two points on the Y axis, and the X-axis equidistant.

第五步是根据坐标系统提取出具有动态尺寸的子图像1。 The fifth step is to extract a sub-image with dynamic size based on the coordinate system. 子图像1的尺寸和位置基于两个点(x1,y1)和(x2,y2)之间的欧几里得距离(D)。 Euclidean distance and position of a sub-image size based on two points (x1, y1) and (x2, y2) between the (D). 坐标系统中表示子图像1的角的点4、5、6、7分别是(0.25D,0.5D)、(1.25D,0.5D)、(0.25D,-0.5D)和(1.25D,-0.5D)。 Coordinate system represents a sub-image corner points 4,5,6,7 respectively (0.25D, 0.5D), (1.25D, 0.5D), (0.25D, -0.5D) and (1.25D, - 0.5D). 因此,子图像1是每条边均等于欧几里得距离且关于Y轴直线3对称的正方形。 Thus, each edge is a sub-image are equal to the Euclidean distance and the straight line 3 on the Y-axis symmetrical square. 因为子图像依赖于手的特征(手指之间的区域),其对于每个个体的手而言是可重复的。 Because sub-images depends on the characteristics of the hand (the area between the fingers), which is repeatable for each individual hand terms.

图7示出了坐标系统的x和y轴2、3以及重叠在图3的原始图像上的子图像1。 Figure 7 illustrates a coordinate system of x and y-axis in Fig. 2, 3 and superimposed on the original image 3 of the sub-image 1.

第六步是利用双线性插值对子图像1进行提取并规一化为标准尺寸。 The sixth step is the use of a bilinear interpolation image pairs were extracted and normalized to a standard size. 图8示出了所提取出的规一化子图像1。 Figure 8 shows the extracted normalized sublayer image 1.

在获得手掌子图像1时,进行本方法的下一部分。 When obtaining palm sub-image 1, the next part of the process.

2)分析图像的皮肤纹理循环伽柏滤波器是用于纹理分析的有效工具,并具有以下一般形式:G(x,y,θ,u,σ)=12πσ2exp{-x2+y22σ2}exp{2πi(uxcosθ+uysinθ)}---(1)]]>其中i=-1;]]>u是正弦波的频率;θ控制函数的方向,以及σ是高斯包络的标准偏差。 2) Analysis of the image texture of the skin circulation Gabor filter is an effective tool for texture analysis are used, and has the following general form: G (x, y, & theta;, u, & sigma;) = 12 & pi; & sigma; 2exp {-x2 + y22 & sigma; 2} exp {2 & pi; i (uxcos & theta; + uysin & theta;)} --- (1)]]> where i = -1;]]> u is a sine wave frequency; θ control function of the direction, and σ is the standard deviation of the Gaussian envelope. 伽柏滤波器广泛地用在纹理分析中,因此,本领域的普通技术人员将熟悉其针对这种目的的应用。 Gabor filter is widely used in texture analysis, therefore, one of ordinary skill in the art will be familiar with its use for this purpose.

为了使纹理分析对图像亮度的变换更为稳定,通过应用以下公式,将离散伽柏滤波器G[x,y,θ,u,σ]变为零DC:G~[x,y,θ,u,σ]=G[x,y,θ,u,σ]-Σi=-nnΣj=-nnG[i,j,θ,u,σ](2n+1)2---(2)]]>其中(2n+1)2是滤波器的尺寸。 In order to make the texture analysis of image brightness conversion is more stable, by applying the following formula, the discrete Gabor filter G [x, y, θ, u, σ] becomes zero DC: G ~ [x, y, & theta; , u, & sigma;] = G [x, y, & theta;, u, & sigma;] - & Sigma; i = -nn & Sigma; j = -nnG [i, j, & theta;, u, & sigma;] (2n + 1 ) 2 --- (2)]]> where (2n + 1) 2 is the filter size. 事实上,因为奇对称,伽柏滤波器的虚部自动具有零DC。 In fact, since the odd symmetry, the imaginary part of Gabor filter automatically have zero DC. 调整后的伽柏滤波器的用途是对预处理图像进行滤波。 Adjusted Gabor filter is to filter the image preprocessing. 然后,通过以下不等式,对相位信息进行编码:br=1如果Re(Σy=-nnΣx=-nnG~[x,y,θ,u,σ]I(x+x0,y+y0))≥0,---(3)]]>br=0如果Re(Σy=-nnΣx=-nnG~[x,y,θ,u,σ]I(x+x0,y+y0))<0,---(4)]]>bi=1如果Im(Σy=-nnΣx=-nnG~[x,y,θ,u,σ]I(x+xo,y+yo))≥0,---(5)]]>bi=0如果Im(Σy=-nnΣx=-nnG~[x,y,θ,u,σ]I(x+xO,y+yO))<0,---(6)]]>其中I(x,y)是预处理图像,以及(x0,y0)是滤波中心。 Then, by the following inequality, the phase information is encoded: br = 1 if Re (& Sigma; y = -nn & Sigma; x = -nnG ~ [x, y, & theta;, u, & sigma;] I (x + x0, y + y0)) & GreaterEqual; 0, --- (3)]]> br = 0 if Re (& Sigma; y = -nn & Sigma; x = -nnG ~ [x, y, & theta;, u, & sigma;] I ( x + x0, y + y0)) & lt; 0, --- (4)]]> bi = 1 if Im (& Sigma; y = -nn & Sigma; x = -nnG ~ [x, y, & theta;, u, & sigma;] I (x + xo, y + yo)) & GreaterEqual; 0, --- (5)]]> bi = 0 if Im (& Sigma; y = -nn & Sigma; x = -nnG ~ [x, y, & theta;, u, & sigma;] I (x + xO, y + yO)) & lt; 0, --- (6)]]> where I (x, y) is an image pre-processing, and (x0, y0) is filtered center.

参照图9和图10,由于可以预期一些用户会不正确地放置他们的手,一些非掌纹像素将包含在手掌子图像中。 Referring to FIGS. 9 and 10, some users can be expected due to incorrectly place their hands, some non palm palm sub-pixel included in the image. 产生了掩模以指出非掌纹像素的位置。 Produced a mask to indicate the position of non-palm pixel. 因为可以认为图像源是半封闭环境,非掌纹像素来自图像背景的黑色边界。 Image source can be considered as a semi-closed environment, non-palm-pixel black border from the image background. 因此,使用阈值来分割非掌纹像素。 Thus, using a threshold to separate the non-palmprint pixels. 典型地,包括掩模和掌纹特征的特征尺寸为384字节。 Typically, features include masks and palmprint feature size is 384 bytes.

图11示出了预处理图像,图12示出了相应纹理特征的实部,图13示出了相应纹理特征的虚部,以及图14示出了相应掩模。 Figure 11 shows the image pre-processing, FIG. 12 shows the real part of the corresponding texture feature, Figure 13 shows the corresponding texture feature of the imaginary part, and Figure 14 shows the corresponding mask.

可以在以下两个公开文件中找到将伽柏滤波器用于纹理分析的有益讨论。 You can find a useful discussion Gabor filters for texture analysis in two public documents. A.Jain和G.Healey发表在IEEE Transactions on ImageProcessing、1998年第7卷第1号、第124~128页上的、题为“Amultiscale representation including opponent color features fortexture recognition”的文章。 A.Jain and G.Healey published in the IEEE Transactions on ImageProcessing, 1998 Vol. 7, No. 1, on page 124 - 128, entitled "Amultiscale representation including opponent color features fortexture recognition" of the article. 以及D.Dunn和WEHiggins发表在IEEE Transactions on Image Processing、1995年第4卷第4号、第947~964上的题为“Optimal Gabor filters for texturesegmentation”的文章。 And D.Dunn and WEHiggins published in the IEEE Transactions on Image Processing, 1995 Vol. 4 No. 4, 947 - 964, entitled on "Optimal Gabor filters for texturesegmentation" article.

3)掌纹匹配将实部和虚部特征表示为矢量,将其与所存储的掌纹数据的矢量进行比较。 3) matching the palm portion of the real and imaginary characteristics is represented as a vector, which vector is compared with the stored data of the palm. 掌纹匹配基于规一化的汉明距离。 Palmprint matching based on a normalized Hamming distance. 例如,P和Q是两个掌纹特征矩阵,规一化的汉明距离可以描述为:D0=Σi=1NΣj=1NPM(i,j)∩QM(i,j)((PR(i,j)⊗QR(i,j)+PI(i,j)⊗QI(i,j)))2Σi=1NΣj=1NPM(i,j)∩QM(i,j),---(7)]]>其中PR(QR)、PI(QI)和PM(QM)分别是P(Q)的实部、虚部和掩模;当且仅当两个比特PR(I)(i,j),等于QR(I)(i,j)时,布尔操作“”的结果等于零;∩表示与操作,以及特征矩阵的尺寸为NN。 For example, P and Q are two palmprint feature matrix, the normalized Hamming distance may be described as: D0 = & Sigma; i = 1N & Sigma; j = 1NPM (i, j) & cap; QM (i, j) (( PR (i, j) & CircleTimes; QR (i, j) + PI (i, j) & CircleTimes; QI (i, j))) 2 & Sigma; i = 1N & Sigma; j = 1NPM (i, j) & cap; QM (i , j), --- (7)]]> where PR (QR), PI (QI) and PM (QM) are P (Q) of the real part and the imaginary part of the mask; iff two bit PR (I) (i, j), equal to QR (I) (i, j) when, "" Boolean operation result is equal to zero; ∩ represents the size and operation, and the feature matrix is N N. 应当注意,D。 It should be noted, D. 在1和0之间。 Between 1 and 0. 对于完全匹配,匹配得分为零。 For an exact match, the matching score is zero. 因为不完全的预处理,需要对特征进行垂直和水平转换,然后再进行匹配。 Because of incomplete pretreatment, the need for vertical and horizontal conversion characteristics, and then the match. 于是,垂直和水平转换的范围是-2到2。 Thus, vertical and horizontal conversion range of -2 to 2. 将通过转换匹配而获得的D。 D. The match is obtained by converting 的最小值作为最终匹配得分。 The minimum score as the final match.

以下的实验结果描述了本发明系统的有效性。 The following experimental results described the effectiveness of the inventive system.

利用掌纹扫描仪从154个对象收集掌纹图像。 Use palm scanners to collect 154 objects from the palmprint image. 大约65%的对象是男性。 About 65% of the target is male. 对象的年龄分布如表1所示。 Age distribution shown in Table 1.

每个对象提供两组图像。 Each object provides two sets of images. 每组包含左手掌的10幅图像和右手掌的10幅图像。 Each group contains 10 images of 10 images and left hand palm of. 总共,每个对象提供40幅图像,以创建包含来自于308个不同手掌的6191幅图像的图像数据库。 In total, each object provides 40 images to create a contained from 308 different palms of 6191 images of image database. 从每个对象收集第一和第二组图像之间的平均时间间隔为57天。 The average time interval from each object to collect between the first and second set of images was 57 days. 最大和最小时间间隔分别为90和4天。 The maximum and minimum intervals were 90 and 4 days. 在完成第一次收集之后,改变光源,并将焦点调节到CCD摄像机上,从而通过两个不同的掌纹扫描仪来模拟图像收集。 After the completion of the first collection, change the light source, and adjusted to focus the CCD camera, thereby to simulate an image was collected by two different palm scanner. 图15和16示出了在针对一个对象的第一和第二组中所捕获的相应手图像。 Figures 15 and 16 are shown in the respective first and second sets in the hand image captured against an object. 所收集的图像具有两种尺寸:384284和768568。 The collected image has two dimensions: 384 284 and 768 568. 将较大的图像的尺寸调整为384284;因此,以下实验中的所有测试图像的尺寸为384284,分辨率为75dpi。 The larger the image resizing for 384 284; therefore, the following experiment all test image size is 384 284, a resolution of 75dpi.

为了获得掌纹系统的验证精度,将每个掌纹图像与数据库中的所有掌纹图像进行匹配。 In order to get the system to verify the accuracy of the palm, and each palmprint image database of all palmprint images to match. 将匹配标记为来自相同对象的相同手掌的两个掌纹图像的正确匹配。 The match marked the correct match between two palmprint images of the same object from the same palm. 比较总数为19161145。 Compare the total number of 19,161,145. 正确匹配数为59176。 Correct matching number 59176.

分别通过正确和不正确匹配来估计真实的和冒名顶替的概率分布示于图17。 Respectively, through correct and incorrect estimate the probability of matching the real and the impostor distribution is shown in Figure 17. 图18示出了相应的接受操作曲线(ROC),是针对所有可能操作点的真实接受比率对错误接受比例的曲线。 Figure 18 shows the corresponding receiving operation curve (ROC), the true acceptance rate for all the possible operating points of the false acceptance ratio curve. 根据图18,可以估计根据本发明的方法可以以96%的真实接受比率和0.1%的错误接受比率进行操作;相应的阈值为0.35。 According to Figure 18, can be estimated may be true acceptance rate of 96% and 0.1% false acceptance rate of the method according to the present invention operates; corresponding threshold value 0.35. 此结果可以与现有掌纹解决方案和包括手几何学和指纹验证在内的其他基于手的生物统计学技术相比。 This results with existing solutions and include palm hand geometry and fingerprint verification, including compared to other biometric technology based hand.

根据本发明的方法利用低分辨率图像,并具有较低的计算成本。 The use of low-resolution images according to the method of the invention, and has a lower computational cost. 验证精度可以与使用高分辨率图像的高性能方法相比。 Verify the accuracy of high-resolution images can be compared with a high-performance method.

此方法可以用于接入控制、ATM和多种安全系统。 This method can be used for access control, ATM, and a variety of security systems.

图19和20示出了根据本发明的掌纹图像捕获设备。 Figures 19 and 20 shows a palmprint image capture according to the present invention apparatus. 所述设备包括外壳1,具有平坦的上表面2,将手放置在其上,手掌向下,以便捕获掌纹图像。 The apparatus includes a housing 1, having a flat upper surface 2, the hand placed thereon, palm down, in order to capture palm print images. 表面2是不透明的,具有通过其捕获图像的窗口8。 Surface 2 is opaque, having captured image through its window 8. 在优选实施例中,窗口8包括玻璃板。 In the preferred embodiment, the window 8 includes glass. 在可选实施例中,窗口8可以包含其他透明的遮盖物、透镜或凹口(即,开放窗口)。 In an alternative embodiment, the window 8 may contain other transparent covering, lens or notches (i.e., open the window).

将如电荷耦合器件(CCD)4等图像传感器安装在外壳1中。 The charge-coupled device (CCD) 4 and so the image sensor is mounted in the housing 1. 将透镜5旋紧在CCD上。 Tighten the lens 5 on the CCD. 透镜5的孔径朝向表面2中的窗口8。 5 toward the surface of the aperture of the lens 2 in the window 8.

安装环形光源6,围绕透镜5,以照亮窗口8中的图像。 Mounting ring light 6 around the lens 5, to illuminate the window 8 in the image. 安装臂7支撑环形光源6,并使用螺丝钉9将CCD牢固地安装到安装臂7上。 An annular mounting arm 7 supporting the light source 6, and 9 using screws CCD securely mounted to the mounting arm 7. 可以通过从透镜5到CCD 4的此光学平面形成掌纹图像,然后将数字化的图像数据传送到如个人计算机(未示出)等外部处理器,以便进行处理和操作。 By forming palmprint image from the lens 5 to the optical plane of CCD 4, and then transferred to the digitized image data such as a personal computer (not shown) and other external processor, for processing and operation.

参照图21,示出了通过图19中的截面A-A'的透镜5和光源6的平面图。 Referring to Figure 21, shows a cross section through FIG. 19 A-A 'of the lens 5 and 6 are a plan view of a light source. 透镜5位于环形光源6的中心。 Lens 5 is located at the center of the annular light source 6. 将透镜5安装在CCD 4的顶部。 The lens 5 is mounted on top of the CCD 4 in.

与表面2中的窗口8相邻的是多个凸起,为栓3的形式,用于将手正确地定位在表面2上,使手掌区域位于窗口8的上方。 2 with the surface of the window 8 is adjacent to a plurality of projections, in the form of plug 3 for correct positioning of the hand on the surface 2, so that the palm region 8 is located above the window. 在使用时,人们将手放在表面2上,使栓3位于拇指和其他手指之间。 In use, it will hand on surface 2, so that bolt 3 is located between the thumb and fingers. 这样确保手正确地放置在设备上,以便通过窗口8来捕获手掌的最佳区域。 This ensures that the right hand placed on the device, so to capture the best area of the palm through the window 8.

图22示出了通过窗口8捕获的目标手掌区域的图像。 Figure 22 shows a window 8 by capturing the palm region of the target image. 显而易见的是,使用具有目标窗口8的不透明表面2确保能够相应地获得手掌上感兴趣的区域。 Be apparent that the use of an opaque surface having a target window 8 2 ensure that access to the region of interest corresponding to the palm. 个人计算机从CCD 4获得此图像,以进行进一步的处理。 This image is obtained from the personal computer CCD 4, for further processing.

由所述设备获得的掌纹适合于用在生物统计学识别中。 Palm obtained by the device suitable for use in biometric recognition. 可以获得掌纹的特征和特性,然后,与数据库记录进行比较,以识别个体。 Palmprint features and characteristics can be obtained, and then, comparing with the database records to identify the individual. 多种技术可以用于确定图像中的手掌的特性。 Various techniques can be used to determine the characteristics of the image of the palm. 一种适合的技术是纹理分析。 One suitable technique is texture analysis. 纹理分析是合适的,因为其能够基于低分辨率图像给出较高的精度。 Texture analysis is suitable because it can give a higher precision based on the low resolution image.

所描述的实施例使用了CCD图像传感器。 The described embodiments use a CCD image sensor. 在可选实施例中,使用互补金属氧化物半导体(CMOS)传感器。 In an alternative embodiment, the use of complementary metal oxide semiconductor (CMOS) sensor. CMOS传感器以更低的成本产生更低的分辨率。 CMOS sensors to produce at a lower cost lower resolution. 但是,如果使用纹理分析,则能够对其加以改善。 However, if you use texture analysis, it is possible to improve them.

在优选实施例中,与窗口8相邻的凸起是栓3。 In a preferred embodiment, the window 8 is adjacent raised bolt 3. 在可选实施例中,以其中能够手掌向下地放置手的凹陷或凹形来形成具有窗口8的表面2。 In an alternative embodiment, in which the hand is placed palm down can be recessed or concave surface 2 is formed with a window 8.

该设备可以用于捕获用在上述方法中的图像。 The device can be used in the above method is used to capture the image.

在前述描述中,以相同的整数或元件来表示已知等价物,如这里单独声明的那样,也包括这些等价物。 In the foregoing description, the same elements used to represent integer or known equivalents, such as where a separate statement, but also includes these equivalents.

已经对本发明的实施例进行了描述,但是应当理解,可以进行改变、改进或修改,而并不偏离本发明的精神或所附权利要求的范围。 Embodiments of the present invention have been described, it is to be understood that changes, improvements or modifications without departing from the spirit or scope of the appended claims of the present invention is required.

Referenced by
Citing PatentFiling datePublication dateApplicantTitle
CN101093626B27 Jul 20079 Feb 2011哈尔滨工业大学Palm print cipher key system
CN101809581B *9 Sep 200810 Dec 2014苹果公司Embedded authentication systems in an electronic device
CN103955674A *30 Apr 201430 Jul 2014广东瑞德智能科技股份有限公司Palm print image acquisition device and palm print image positioning and segmenting method
CN103955674B *30 Apr 201410 May 2017广东瑞德智能科技股份有限公司掌纹图像采集装置及掌纹图像定位与分割方法
US87827759 Sep 200815 Jul 2014Apple Inc.Embedded authentication systems in an electronic device
US89435809 Sep 200827 Jan 2015Apple Inc.Embedded authentication systems in an electronic device
US903816727 Dec 201319 May 2015Apple Inc.Embedded authentication systems in an electronic device
US912860118 Mar 20158 Sep 2015Apple Inc.Embedded authentication systems in an electronic device
US913489627 Dec 201315 Sep 2015Apple Inc.Embedded authentication systems in an electronic device
US925079527 Dec 20132 Feb 2016Apple Inc.Embedded authentication systems in an electronic device
US92746471 Oct 20151 Mar 2016Apple Inc.Embedded authentication systems in an electronic device
US93046245 Sep 20145 Apr 2016Apple Inc.Embedded authentication systems in an electronic device
US932977120 Jun 20143 May 2016Apple IncEmbedded authentication systems in an electronic device
US93426745 Mar 201517 May 2016Apple Inc.Man-machine interface for controlling access to electronic devices
US94955315 Feb 201615 Nov 2016Apple Inc.Embedded authentication systems in an electronic device
US951977127 Dec 201313 Dec 2016Apple Inc.Embedded authentication systems in an electronic device
US974083230 Aug 201322 Aug 2017Apple Inc.Method, apparatus and system for access mode control of a device
Classifications
International ClassificationG06K9/00, A61B5/117
Cooperative ClassificationG06K9/00067, G06K9/00362, A61B5/117
European ClassificationG06K9/00A2, G06K9/00H
Legal Events
DateCodeEventDescription
19 Oct 2005C06Publication
14 Dec 2005C10Request of examination as to substance
9 Apr 2008C14Granted