CN1170912A - Character recognition result display method, character recognition system and information recording medium - Google Patents

Character recognition result display method, character recognition system and information recording medium Download PDF

Info

Publication number
CN1170912A
CN1170912A CN 97113677 CN97113677A CN1170912A CN 1170912 A CN1170912 A CN 1170912A CN 97113677 CN97113677 CN 97113677 CN 97113677 A CN97113677 A CN 97113677A CN 1170912 A CN1170912 A CN 1170912A
Authority
CN
China
Prior art keywords
character
identification result
document image
datum line
image
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN 97113677
Other languages
Chinese (zh)
Other versions
CN1103088C (en
Inventor
工藤奈保子
金子馨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ricoh Co Ltd
Original Assignee
Ricoh Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ricoh Co Ltd filed Critical Ricoh Co Ltd
Publication of CN1170912A publication Critical patent/CN1170912A/en
Application granted granted Critical
Publication of CN1103088C publication Critical patent/CN1103088C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Abstract

A method of displaying character identification results is used to identify the characters of an inputted document image, displays the image of the obtained character identification result and the inputted document image, and judges if character identification is correctly performed, when the character identification is faulty, the characters are modified into correct characters, and from here, the characters of the character identification result, which become the object judging the character identification result, and the characters in the inputted document image, which become the character identification source of the characters of the character identification result, are adjacently displayed.

Description

Character recognition result display method, character recognition system and carrier
The present invention relates to carrying out from the character image that cuts out of input document image that character recognition is handled and the character identification result that obtains being presented at character recognition result display method, character recognition system and carrier on the picture.
Usually in character recognition system, read in documents such as original copy with scanner etc., it as input document image, is cut out character image from this input document image then, obtain to carry out the character identification result of character recognition processing.In this system, for work such as the affirmation that makes character identification result, correction are carried out easily, in the past, shown in for example special fair 7-72903 number, both are presented at function on the picture to have the character identification result that will be obtained and original input document image thereof.
More particularly, as shown in Figure 1, in the character recognition device shown in the fair 7-72903 of spy number, will be presented on the display part side by side simultaneously by the input document image of double quantification and character identification result image.Promptly, in example shown in Figure 1, left-half at the display part picture shows input document image, right half part at the display part picture shows and the input document image corresponding characters recognition result image that shows in left-half, can also go up volume as required, so as to make the input document image that shows in left-half and the character identification result image that shows at right half part in correspondence with each other.Specifically, under state shown in Figure 1, if the input document image that will show in left-half scrolling 2 row upward then as shown in Figure 2, also can make with its interlock and upward scrolling of the character identification result image that shows at right half part.
Like this, in above-mentioned existing character recognition device, input document image and character identification result image both sides can both show, roll up by making accordingly on the opposing party's the image with a side image wherein, the operator can be in the contrast of enterprising line character recognition result of the display frame of display part and original copy.
; in above-mentioned existing character recognition system; in the affirmation of carrying out character identification result, when revising; be presented at character identification result and original input document image thereof in the display frame accordingly; in addition; can make them go up volume mutually linkedly, there is the problem of the corresponding relation that does not see character identification result image and input document image in such image.In other words, when the operator contrasts character identification result image in display frame with input document image, all to force to carry out about sight line to move significantly at every turn, increase fatigue strength, the problem that exists work efficiency to descend.In addition, carry out the affirmation of character identification result, when revising, the operator must watch whole input document images, particularly in the affirmation of carrying out a large amount of character identification results, when revising, the problem that exists work efficiency to descend.
The object of the present invention is to provide a kind of when work such as the affirmation of carrying out character identification result, correction, the character recognition result display method that can increase work efficiency significantly, character recognition system and carrier.
In order to achieve the above object, the 1st aspect of the present invention be a kind of to the input document image carry out character recognition, the image and the input document image of the character identification result that obtains are presented on the picture, and whether the identification of judging character is correct, when the identification of concluding character is wrong, be modified to the character recognition result display method that correct character is used
It is characterized in that: be adjacent to show from the character of the character identification result that will become the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
In addition, the 2nd aspect of the present invention is a kind ofly given input document image is carried out character recognition to handle, the character identification result that obtains is presented at character recognition result display method in the display frame, it is characterized in that: the datum line during with the character display recognition result is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
In addition, the 3rd aspect of the present invention is a kind ofly given input document image is carried out character recognition to handle, the character identification result that obtains is presented at character recognition result display method in the display frame, it is characterized in that: in character recognition is handled, calculate the accuracy of character identification result with character identification result, the character group of character identification result is presented in the display frame, when cursor being presented at desire wherein and confirming from the character image of the character of the character identification result here, cross the character of (skipping) accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
In addition, the document image of the image of recognition result of the character recognition device of character recognition and character display identification and input is carried out in the 4th aspect of the present invention by the document image input media of input document image, to the document image of input image display apparatus constitutes
It is characterized in that: image display apparatus will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
In addition, the 5th aspect of the present invention is characterised in that: the image input device that has input document image, document image to this input carries out the character recognition treating apparatus that character recognition is handled, the character identification result of this character recognition treating apparatus is presented at display device in the display frame, and the display control unit that when being presented at this character identification result on the display device, shows control, the datum line of display control unit during with the character display recognition result is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
In addition, the 6th aspect of the present invention is characterised in that: the invention described above the 5th aspect character recognition system in comprise setting device, be used for before the input document, be to erect to write document or write across the page document according to the input document image of importing by operator's operation in advance, set datum line along laterally or longitudinally showing.
In addition, the 7th aspect of the present invention is characterised in that: the invention described above the 5th aspect character recognition system in also be provided with the datum line setting device that the condition of setting datum line is used, display control unit shows datum line according to the condition of the datum line of being set by the datum line setting device.
In addition, the 8th aspect of the present invention is characterised in that: the invention described above the 7th aspect character recognition system in, the condition of datum line is such, is to erect to write document or write across the page document according to input document image promptly, in advance the datum line edge laterally or is longitudinally shown.
In addition, the 9th aspect of the present invention is characterised in that: the invention described above the 5th aspect character recognition system in, when the picture area of character display recognition result image showed as window respectively with the picture area that shows input document image, the lap that datum line is used as both sides' window frame showed.
In addition, the 10th aspect of the present invention is characterised in that: the image input device that has input document image, document image to this input carries out the character recognition treating apparatus that character recognition is handled, the character identification result of this character recognition treating apparatus is presented at display device in the display frame, obtain the accuracy calculation element of accuracy of the character identification result of character recognition treating apparatus, and the display control unit that when being presented at this character identification result on the display device, shows control, the character group of display control unit character display recognition result, when cursor being presented at desire wherein and confirming from the character image of the character of the character identification result here, cross the character of accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
In addition, the 11st aspect of the present invention by stage of input document image a), the stage c) that the document image of input carried out the document image of the image of recognition result of the stage b) of character recognition and character display identification and input constitutes.
It is characterized in that: in image shows stage c), will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
In addition, the 12nd aspect of the present invention is the software program that computing machine is used, be used for input document image is carried out character recognition, the image and the input document image of the character identification result that obtains are presented on the picture, and whether the identification of judging character is correct, when the identification of concluding character is wrong, be modified to correct character, the software program that this computing machine is used is recorded on the carrier, it is characterized in that: will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
In addition, the 13rd aspect of the present invention is the software program that computing machine is used, the stage that is used for carrying out input document image a), the document image stage b) that carries out character recognition to input, and the stage c) of the document image of the image of the recognition result of character display identification and input, the software program that this computing machine is used is recorded on the carrier, it is characterized in that: in image shows stage c), will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
In addition, the 14th aspect of the present invention is characterised in that: the software program that computing machine is used is recorded on the carrier, the software program that this computing machine is used is used for when the character display recognition result, datum line is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, in addition, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
In addition, the 15th aspect of the present invention is characterised in that: the software program that computing machine is used is recorded on the carrier, the software program that this computing machine is used is used in character recognition is handled, calculate the accuracy of character identification result with character identification result, the character group of character identification result is presented in the display frame, when cursor being presented at desire and confirming from the character image of the character of the character identification result here, cross the character of accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
If adopt above-mentioned the present invention, then when work such as the affirmation of carrying out character identification result, correction, can increase work efficiency significantly.
Particularly the invention described above the 1st aspect, the 4th aspect, the 11st aspect, the 12nd aspect and the 13rd aspect, will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.In addition, the invention described above the 2nd aspect, the 5th aspect, the 9th aspect and the 14th aspect, datum line during with the character display recognition result is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.Therefore, the operator can will become present character identification result easy as can and confirm to revise the character string of the character string of character identification result (text) of the object input document image (character image) original with it and contrast (not needing to make sight line to do very big moving just can contrast) on picture, can find out easy as can become present character identification result confirm to revise places different in the character identification result (text) of object with original input document image (character image) (misidentify the place or the place of generation omission etc.).
In addition, the invention described above the 3rd aspect, the 10th aspect, the 5th aspect, because in character recognition is handled, calculate the accuracy of character identification result with character identification result, character identification result is presented in the display frame, when cursor being presented at desire wherein and confirming from the character image of the character of the character identification result here, cross the character of (skipping) accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value, so in the affirmation of carrying out character identification result, revise when waiting, the operator does not need to watch whole documents, can increase work efficiency.
Other purpose of the present invention and feature will become clearer and more definite by the detailed description of doing with accompanying drawing.
Fig. 1 is the existing figure that shows example of expression input document image and character identification result.
Fig. 2 is the key diagram that the existing upward volume function of explanation is used.
Fig. 3 is the structure illustration of expression character recognition system of the present invention.
Fig. 4 is the hardware configuration illustration of the character recognition system in the presentation graphs 3.
Fig. 5 is the illustration of certain one page document of expression.
Fig. 6 A, Fig. 6 B be expression to the character identification result of document shown in Figure 5 confirm, the figure of the demonstration example of usefulness such as correction.
Fig. 7, Fig. 8 be explanation in display frame, make the character identification result (text) that shows in a side of datum line and the input document image (character image) that shows at the opposite side of datum line often interlock, go up volume or key diagram that the function that moves is used.
Fig. 9 be expression to the character identification result of document shown in Figure 5 confirm, the figure of the demonstration example of usefulness such as correction.
Figure 10 is the process flow diagram that the 1st of expression character recognition result display method of the present invention is handled example.
Figure 11 is the process flow diagram that the 2nd of expression character recognition result display method of the present invention is handled example.
Figure 12, Figure 13, Figure 14 be represent respectively to the character identification result of document shown in Figure 5 confirm, the figure of the demonstration example of usefulness such as correction.
Below, embodiments of the invention are described with reference to the accompanying drawings.Fig. 3 is the structure illustration of expression character recognition system of the present invention.With reference to Fig. 3, this character recognition system has: with the image input section 1 of documents such as original copy as the document image input; The input document image storage part 2 of the input document image that storage is read in by image input section 1; From by cutting the character image that to become the character recognition object the input document image of image input section 1 input, the feature of taking out this character image compares the character recognition handling part 4 that the line character identification of going forward side by side is handled with given dictionary 5; Storage is from the character identification result storage part 6 of the character identification result of character recognition handling part 4; The display part 7 that shows etc.; For work such as the affirmation of carrying out character identification result, corrections, carry out character identification result is presented at the display control unit 8 of the control on the picture of display part 7; And the condition enactment portion 9 that uses of the conditions such as display format of setting character identification result.
Fig. 4 is the hardware configuration illustration of the character recognition system in the presentation graphs 3.With reference to Fig. 4, this character recognition system has: the CPU11 that is totally controlled by the carrying out of realizations such as for example personal computer; The ROM12 of the control program of storage CPU11 etc.; The RAM13 that uses as the workspace of CPU11 etc.; The scanner 14 that documents such as original copy are read in as document image; The external memory 15 of storage input document image file, lexicon file and text; Set the condition such as display format of character identification result and to the character identification result (text) that obtains is confirmed, work such as correction is used display 18 and input media 19.In addition, above-mentioned input document image file is the file of the input document image (or cutting the character image that is used for character recognition of going out from input document image) that read in by scanner 14, and this document for example is that the page or leaf with the input document is that the unit compression forms.Above-mentioned lexicon file is the file of the dictionary used of character recognition.Above-mentioned text is that input document image (character image) is carried out the file that the text (information that is encoded) of the character identification result that obtains is handled in character recognition.
Here, scanner 14 is corresponding to the image input section among Fig. 31, and the external memory 15 of storage input document image file, dictionary and text is corresponding to input document image storage part 2, dictionary 5, character identification result storage part 6 among Fig. 3.In addition, CPU11 has the character recognition handling part 4 among Fig. 3 and the function of display control unit 8.
In addition, function as this character recognition handling part among the CPU11, display control unit etc. for example can be with software package (specifically, carriers such as CD-ROM) form provides, therefore, in example shown in Figure 4, when mount message recording medium 20, be provided with the medium drive 21 that drives it.
In addition, for example can use keyboard, Genius mouse etc. as input media 19, for example by the icon that on the picture of display 18, shows with mouse indication etc., the selection of handling, the beginning of given processing, end indication etc., utilize keyboard or Genius mouse, moving of the enterprising line cursor of picture, can also carry out the volume of going up of picture.
In other words, character recognition system of the present invention can constitute a kind of like this system to be realized, the procedure code that is about to write down in the carrier such as CD-ROM reads in the general calculation machine systems such as having image scanner, display, carries out character recognition and handle in the microprocessor of this general-purpose computing system.At this moment, be not limited to CD-ROM, also can use ROM, RAM, FD etc. as the carrier of storage character recognition handling procedure of the present invention etc.In addition, the input of document image is not limited to use scanner, also can be used as image file and supplies with from the outside.
; in the present invention; display control unit 8 is in the affirmation of carrying out character identification result (text); during work such as correction; datum line during with the character display recognition result is presented on the picture of display part 7 (display 18); the character string that shows the character identification result (text) that becomes present character identification result affirmation correction object in a side of this datum line along datum line; showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image (character image) of the character string of character identification result (text) of object, and at this moment making becomes that present character identification result confirms to revise the character string of character identification result (text) of object and the character string of the input document image (character image) corresponding with it clips datum line (side by side) demonstration relatively mutually.
Fig. 5 represents the example of certain one page document, Fig. 6 A, Fig. 6 B represent respectively to the character identification result of document shown in Figure 5 confirm, the demonstration example of usefulness such as correction.
In the demonstration example of Fig. 6 A, on picture, demonstrate datum line L longitudinally 1, along this datum line L 1,, follow direction (=vertical) and show and become the character string " the fine な り of this day は, " that present character identification result confirms to revise the character identification result (text) of object near its place on its right side, in addition, datum line L along the longitudinal 1, follow direction (=vertical) and show and become above-mentioned character identification result and confirm to revise the character string " this day は fine day な り, " of the corresponding input document image (character image) of the character string of character identification result (text) of object near its place in its left side.
In addition, in the demonstration example of Fig. 6 B, on picture, demonstrate horizontal datum line H 1, along this datum line H 1, side follows direction (=horizontal stroke) and shows that becoming present character identification result confirms the character string of the character identification result (text) of correction object " the fine な り of this day は, " near its place thereon, in addition, and datum line H transversely 1In the place of its downside, follow the character string " the fine な り of this day は, " that direction (=horizontal stroke) shows the input document image (character image) corresponding with the character string of the character identification result (text) that becomes above-mentioned character identification result affirmation correction object near it.
Moreover in display frame, the following setting as shown in Figure 6A, is vertically line direction, perhaps shown in Fig. 6 B, makes line direction be horizontal.For example before the document input of being undertaken by scanner 14, use input media 19 inputs is to erect to write document or the information of the document of writing across the page about the document of being imported to the operator.If the result of this input perpendicular writes document, if then the CPU11 result who longitudinally sets this input of datum line like that as shown in Figure 6A is the document of writing across the page, just shown in Fig. 6 B like that along the lateral set datum line.
In addition, in the demonstration example of Fig. 6 A and Fig. 6 B, show that becoming character identification result confirms that the picture area of the character identification result (text) of correction object is used as window (text confirms to revise window) W 1Set, confirm to revise window W in the text 1In, the character identification result that becomes text of reading and demonstrate in the external memory 15 storage confirms to revise all or part of of character identification result (text) of object, at this moment, make become present character identification result confirm to revise object character identification result (text) character string and datum line is the most approaching shows.
In addition, in the demonstration example of Fig. 6 A and Fig. 6 B, show that the picture area of input document image (character image) is used as window (visual window) W 2Set, in the case, at the visual window W of this picture 2In, read and demonstrate the input document image (character image) of the input document image file of storage in the external memory 15, at this moment, the character string of the input document image (character image) corresponding with the character string of the character identification result (text) that shows near datum line in a side of datum line is shown near datum line.
Like this, when the picture area of character display recognition result (text) and the picture area that shows input document image (character image) by (that is, text confirms to revise window W as window respectively 1, visual window W 2) when setting, can be with datum line L 1Or H 1(that is, text confirms to revise window W as the intersection of both sides' window frame 1Window frame and visual window W 2The intersection of window frame) show.As mentioned above, the operator is before the input document, what input media 19 inputs were used in input is to erect to write document or the information of the document of writing across the page about document, CPU11 is according to this input results, set window (promptly, text confirms to revise window, visual window) how to show, thus the display format that can automatically set datum line (makes datum line resemble L 1Show like that or resemble H 1Show like that).
In addition, in the present invention, display control unit 8 has the character identification result (text) that makes on the picture of display part 7, show in a side of datum line and the input document image (character image) that shows at the opposite side of datum line often interlock, the function that goes up volume or move.Specifically, can carry out such demonstration control, that is, for example under the state of Fig. 6 A, if make at datum line L 1The character identification result (text) that shows of right side 1 row that for example moves right, then as shown in Figure 7, with its interlock, at datum line L 1The input document image (character image) that shows of left side 1 row that also moves right.In addition, can also carry out such demonstration control, that is, for example under the state of Fig. 6 A, if make at datum line L 1The character identification result (text) that shows of right side 2 characters that for example move up, then as shown in Figure 8, with its interlock, at datum line L 1The input document image (character image) that shows of left side 2 characters that also move up.In addition, such character identification result, going up of input document image are rolled up, are moved and can both move in window separately linkedly mutually.
Like this, because the input document image (character image) that has the character identification result (text) that makes on the picture of display part 7, show in a side of datum line and show at the opposite side of datum line is interlock, the function that goes up volume or move often, so also can know from Fig. 7 or Fig. 8, can often make become character string that present character identification result confirms to revise the character string of character identification result (text) of object and the input document image (character image) corresponding with it along datum line and clip mutually datum line relatively (side by side) show.
In addition, display control unit 8 can also show like this, promptly carries out the affirmation of character identification result, when revising, about the character in the present correction, can identify now and it be revised.Specifically, display control unit 8 has such function, promptly in Fig. 6 A institute example, as shown in Figure 9, when the character in " days " are mistaken as " husband " and revise now in character identification result (text) is " husband ", then in the viewing area of character identification result, for example place, character " husband " place is fenced up with rectangle frame, in addition, in the viewing area of input document image, for example will fence up with the place, character image place of " days " of its corresponding characters with red rectangle frame.
In addition, display control unit 8 can come out the low accuracy character in the character identification result (text) and other character recognition in the affirmation of carrying out character identification result (text), when revising.For example, in the viewing area of character identification result (text), with the low accuracy character flip displays in the character identification result (text), in addition, and in the viewing area of input document image (character image), also can be with above-mentioned low accuracy character flip displays.Above-mentioned so-called low accuracy character is meant the low character of accuracy (reliability) by the result of 4 pairs of character recognition of character recognition handling part.
In addition, in the demonstration example of Fig. 6 A, Fig. 6 B etc., the character of input document image (character image) can be pressed the onesize demonstration of character of the document original, adjust the character size of character identification result (text), can press and import the onesize demonstration of character of document image with it.
In addition, in the demonstration example of Fig. 6 A, Fig. 6 B etc., though, as described later, on this picture, can also show for example all images of character identification result affirmation correction object page or leaf at picture on character display recognition result (text) and input document image (character image).In addition, can also confirm that all images of revising the object page or leaf are adjacent to show information such as the correction number of characters in this page, the number of characters of low accuracy, total number of characters with character identification result.
As mentioned above, when showing the character of low accuracy, character recognition handling part 4 is obtained character identification result, obtains the accuracy (reliability) of this character identification result simultaneously.The computing of this accuracy for example can be adopted the spy to open in flat 4-211883 number disclosed method to carry out.
That is, so-called accuracy is which kind of degree the character of the last character identification result of expression can come to, and enough from 0% to 100% the numeric representations of energy perhaps can change into several stages with this numerical value quantum and represent.For example can represent with following A, B, a C3 grade.
The A level: the correct possibility of character identification result is very high.
The B level: the correct possibility of character identification result is low.
The C level: the correct possibility of character identification result is very low.
In character recognition handling part 4,, synthetically determine accuracy according in order to obtain the information that processing that last character identification result passes through a plurality of stages obtains.For example, the information of the representation language correction result that will criticize information which kind of rule the expression of joining poor, the evaluation of estimate when obtain definite qualified handled in qualified selection of handling the 1st candidate's evaluation of estimate obtain or the 1st and the 2nd candidate's evaluation of estimate, obtaining from rule treatments can revise with from pattern and obtain from Language Processing puts together, according to these information, and utilize for example probability theory of De Mupusite-Sa Fo (Dempster and Shafer), synthetically accuracy of judgement degree.
Determining of this accuracy, to put together in the lump in the processing stage information that obtains before this processing stage of in the end and determine, the perhaps information that obtains according to reason stage throughout, obtain candidate's accuracy, in view of the above, before arriving final processing stage,, determine by being updated in the operation of the accuracy of trying to achieve before the pretreatment stage repeatedly.
Then, according to this accuracy, when carrying out the demonstration of character identification result, visual conditions such as color by changing character or briliancy, perhaps with the character of character identification result character display or symbol accordingly, system operator can easily be recognized the accuracy of character identification result, can be rapidly and find the character of need revising reliably, can carry out this correction work expeditiously.
Like this, when calculating the accuracy of character recognition with character identification result, this result who calculates can followingly utilize.Display control unit 8 can be in the affirmation of carrying out character identification result (text), when revising, cursor is presented in the character identification result (text) that shows on the picture of display part 7 (display 18) or/and the desire in the input document image (character image) (for example this character being fenced up as mentioned above) on the character of acknowledge character recognition result from here on rectangle frame.At this moment, display control unit 8 can make accuracy that cursor crosses (skipping) character recognition character more than given threshold value (for example 90%), and is presented on the character of low accuracy.That is, the character of the accuracy that can take out its character recognition in the character of character identification result (text) below given threshold value (for example 90%), and cursor is presented on this character.
In addition, the function that cursor only is presented on the character of such low accuracy also can realize in following picture, promptly for example can be at Fig. 6 A, the side at datum line shown in Fig. 6 B shows along datum line becomes the character string that present character identification result is confirmed the character identification result (text) of correction object, show along datum line with the present character identification result that becomes that a side at datum line shows at the opposite side of datum line and to confirm to realize in the display frame of character string of the input document image (character image) that the character string of character identification result (text) of correction object is corresponding, perhaps in a picture of character display recognition result (text), realize, perhaps in the picture that only shows input document image (character image), realize.
In addition, the threshold value of accuracy can be set by condition enactment portion 9 among Fig. 3 or the input media 19 among Fig. 4.
Figure 10 is the process flow diagram that the 1st of expression character recognition result display method of the present invention is handled example.With reference to Figure 10, handle in the example the 1st, during the character display recognition result, datum line during with the character display recognition result is presented at (step S1) in the display frame, showing along datum line and the most close datum line in a side of this datum line becomes the character string (step S2) that present character identification result confirms to revise the character identification result of object, and showing with a side at datum line shows along datum line and the most close datum line at the opposite side of this datum line becomes the character string (going on foot S3) that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object.Thereby can make become present character identification result confirm to revise the character string of the character string of character identification result of object and the input document image corresponding with it clip mutually datum line relatively (side by side) show.
In addition, Figure 11 is the process flow diagram that the 2nd of expression character recognition result display method of the present invention is handled example.With reference to Figure 11, handle in the example the 2nd, in character recognition is handled, calculate the accuracy (step S11) of character identification result with character identification result.Secondly, the character identification result that obtains is presented at (step S12) in the display frame, but searching character recognition result is at this moment checked the character (step S13) of accuracy below given threshold value.Then, judge whether the character (step S14) of accuracy below given threshold value, when the character of accuracy below given threshold value, arrived accuracy before the character below the given threshold value, make cursor cross (skipping) character, (go on foot S15) and be presented on the character of accuracy less than this given threshold value.In other words, make cursor cross the character of (skipping) accuracy more than given threshold value, and be presented at accuracy on this character below given threshold value.So when the affirmation of carrying out character identification result, correction etc., the operator does not need to watch whole documents, can increase work efficiency.
Secondly, the concrete example of the work of treatment of the character recognition system that constitutes like this is described.When character recognition system of the present invention provided with the form of software package (carrier) 20, the operator was installed in this carrier 20 in the medium drive 21, and character recognition system software is for example packed among the RAM13.In this stage, CPU11 just can handle according to the character recognition system software among the RAM13 that for example packs into.
When scanner 14 for example is ADF (original copy automatic supplier), if for example many original copys place scanner 14, just then scanner 14 automatically reads many original copys successively.For purposes of simplicity of explanation, suppose that each original copy is individual single face original copy, therefore 1 original copy, when for example reading n (n 〉=1) original copy, is the document image of the input n of unit page or leaf with the page or leaf corresponding to 1 page.They as input document image file, are deposited in external memory storage 15 by respectively successively.
CPU11 utilizes the dictionary in the lexicon file, with the page or leaf is unit, input document image file to the input document image of the n page or leaf of storage in the external memory 15 carries out the character recognition processing, with per 1 page character identification result as text (information that is encoded), and with them as text, deposit external memory 15 successively in.
The operator can confirm, revise with page or leaf position unit being the character identification result (text) of the n page or leaf of unit storage with the page or leaf as text respectively for example in this stage.When carrying out this affirmation correcting process, the icon of operator by tapping the regulation that shows in the display frame (for example " confirm to revise " etc. icon) comes the initiated innovation adjustment processing program.Confirm adjustment processing program in case started (perhaps as input document image file, text, when storing a plurality of document, if be selected to the document that character identification result confirms to revise object), then as shown in figure 10, showing on the picture: datum line (in example shown in Figure 10, being vertical datum line) L1, the text that shows character identification result (text) usefulness of the document that becomes character identification result affirmation correction object confirms to revise window W1, show the visual window W2 that the input document image corresponding with the character identification result (text) that becomes character identification result affirmation correction object is used, the character display recognition result is confirmed all window W3 of page or leaf of whole visual usefulness of correction object page or leaf, and be presented at number of characters in the character identification result of revising in this affirmation correcting process (text), the number of characters of low accuracy, the information display window W4 of total number of characters (all, become present character identification result and confirm to revise number in the page or leaf of object) usefulness.
Secondly, CPU11 reads the input document image file that character identification result confirms to revise all images of object page or leaf, whole images of this page are presented on page all window W3, in addition, read the character identification result (text) of confirming the 1 line character number that the beginning of correction object page or leaf begins from character identification result from given text, and be presented on the text affirmation correction window W1.At this moment, as shown in figure 12, show into the character string of character identification result (text) of confirming to revise 1 line character number of object into present character identification result.(in illustrated example,,, and being adjacent to show) with datum line L1 along datum line L1 on the right side of datum line L1.In addition, CPU11 reads the corresponding input document image (character image) of character string with the character identification result (text) that is adjacent to show on the right side of datum line L1, with datum line L1 from this input document image file, under situation with its compression back storage, it is launched, for example be presented on the visual window W2 according to the size identical with the character of the document of importing by scanner 14.At this moment, as shown in figure 12, the character string of the input document image (character image) corresponding with the character string of the character identification result (text) that is adjacent to show on the right side of datum line L1, with datum line L1 is displayed on the left side of datum line L1, and, be adjacent to show with datum line L1 along datum line L1.That is, become present character identification result confirm to revise the character string of the character string of character identification result (text) of object and the input document image (character image) corresponding with it clip mutually datum line relatively (side by side) show.
Therefore, the operator can will become present character identification result easy as can and confirm to revise the character string of the character string of character identification result (text) of the object input document image (character image) original with it and contrast (not needing to make sight line to do very big moving just can contrast) on picture, can find out easy as can become present character identification result confirm to revise places different in the character identification result (text) of object with original input document image (character image) (misidentify the place or the place of generation omission etc.).
Now, when becoming character identification result confirm to revise for example found out misidentify in the character identification result (text) of object character the time, the operator is presented at this place, character place with cursor.At this moment, as shown in figure 13, the place, this character place with character identification result fences up with rectangle frame, and in addition, the place, corresponding characters image place in the input document image is also with the rectangle frame demonstration that fences up.Under this state, the operator for example imports the correct pronunciation (being " か ぃ " example shown in Figure 13) of this character from keyboard.In view of the above, as shown in figure 13, on picture, the correction candidate characters of this character is displayed on candidate characters and selects on the window W5, the operator is with candidate characters selector button for example etc., from revise candidate characters, select correct character, can be with the misidentify in the character identification result (text) character replacement become correct character.In addition, when for example having found out omission in the character identification result (text) that is becoming character identification result affirmation correction object, the operator is presented at this place with cursor, for example default is become the insertion pattern, for example, can be inserted into the place that omission is arranged in the character identification result (text) from the character of keyboard misinput.
So, present moving cursor, to the text that is presented at 1 line character number on the picture carry out misidentify character or the correction of omission.In addition, when confirm to be presented at the character confirming to revise on the window W1 identical with character on being presented at visual window W2 after, the operator operates movable button (for example scroll key) etc., the character identification result (text) of 1 line character number under for example from text, reading, and be presented on the text affirmation correction window W1.At this moment, the character string of the character identification result (text) of 1 line character number shows (in illustrated example, on the right side of datum line L1, and along datum line L1, being adjacent to show with datum line L1) as shown in Figure 14 down.In addition, CPU11 reads the corresponding input document image (character image) of character string with the character identification result (text) that is adjacent to show on the right side of datum line L1, with datum line L1 from input document image file, under situation with its compression back storage, it is launched, for example be presented on the visual window W2 according to the size identical with the character of the document of importing by scanner 14.At this moment, as shown in figure 14, the character string of the input document image (character image) corresponding with the character string of the character identification result (text) that is adjacent to show on the right side of datum line L1, with datum line L1 is displayed on the left side of datum line L1, and, be adjacent to show with datum line L1 along datum line L1.That is, become character identification result confirm to revise the character string of the character string of character identification result (text) of object and the input document image (character image) corresponding with it clip mutually datum line relatively (side by side) show.
Therefore, the operator can also will descend character identification result (text) the input document image (character image) original with it of 1 row to contrast on picture easy as can, can find out easy as can the place different with original input document image (misidentify the place or the place of take place omitting etc.), can confirm, revise with method same as described above.
In addition, Figure 12 and Figure 14 are compared as can be known, in this embodiment, when the character string of the character identification result (text) that will descend 1 row is presented at the right side near datum line L1, the character string of the character identification result (text) of preceding 1 row is sidesway 1 row demonstration to the right just, in addition, the character string of the input document image (character image) of preceding 1 row since down the character string of the input document image (character image) of 1 row be presented at the left side near datum line L1 and cancellation from the picture.
In other words, the demonstration of the character string of above-mentioned following 1 line character recognition result can show by confirming to revise mobile to the right respectively 1 row of image visual and that show that shows on the window W1 on visual window W2 at file.
In addition, in above-mentioned processing example, the operator is in the place of needs correction, and Genius mouse by input device 19 etc. comes moving cursor.As mentioned above, cross the character of (skipping) pin-point accuracy and be presented under the situation of the function on the character of low accuracy system being had make cursor, CPU11 automatically moves on the character of low accuracy cursor, makes the operator can confirm or revise the character identification result of low accuracy character.Promptly, CPU11 is for example from the beginning retrieval low accuracy character of 1 page character identification result (text), when detecting the low accuracy character, make cursor automatically move to this place, character place, and cursor is presented on this character, make that the operator confirms, correction etc., if after the operator has operated the key that the affirmation of representing this character, correction etc. finished, cursor just automatically moves to place, following 1 low accuracy character place, and the operator can carry out affirmation, the correction of this character.
Therefore, when the affirmation of carrying out character identification result, correction etc., the operator does not need to watch whole documents, can increase work efficiency.
Handle like this, after 1 page affirmation, revising end, the operator for example operates Page Down keys, just can carry out following 1 page affirmation correcting process with same method.In addition, when returning preceding 1 page, for example page key gets final product before the operation.
Then, after whole pages affirmation correcting process finished, the operator clouted for example conclusion button.Determine affirmation, correcting process thus.
Wait in English in the less language of number of characters making, when carrying out the affirmation correction of character identification result, utilize spelling-checker, can find misidentify effectively character.For example under the situation of English, most of article can be only with 26 character representations in the letter.Different therewith, in the more language of number of characters such as Japanese, Chinese, be difficult to realize the corresponding function of spelling-checker with English etc.For example under the situation of Japanese, assumed name and Chinese character are necessary when publishing an article, the number of words of use far away (figure place difference) more than 26 characters in the alphabet.Therefore, the operator must import document image and character identification result, find misidentify character and the used time increase when revising.If employing the present invention, owing to when carrying out the affirmation correction of character identification result, can reduce the time of operator significantly, so can increase substantially the affirmation correction efficient of character identification result, in addition, can also alleviate the degree of fatigue of operator when carrying out the affirmation correction of character identification result.
In addition, embodiments of the invention are not subjected to the restriction of foregoing, can carry out various variations in the scope of following claim.

Claims (15)

  1. One kind to the input document image carry out character recognition, the image and the input document image of the character identification result that obtains are presented on the picture, and judge whether the identification of character is correct, when the identification of concluding character is wrong, be modified to the character recognition result display method that correct character is used
    It is characterized in that: will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
  2. 2. one kind is carried out character recognition to given input document image and handles, the character identification result that obtains is presented at character recognition result display method in the display frame, it is characterized in that: the datum line during with the character display recognition result is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
  3. 3. one kind is carried out character recognition to given input document image and handles, the character identification result that obtains is presented at character recognition result display method in the display frame, it is characterized in that: in character recognition is handled, calculate the accuracy of character identification result with character identification result, the character group of character identification result is presented in the display frame, when cursor being presented at desire wherein and confirming from the character image of the character of the character identification result here, cross the character of (skipping) accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
  4. 4. character recognition system, it carries out the document image of the image of recognition result of the character recognition device of character recognition and character display identification and input by the document image input media of input document image, to the document image of input image display apparatus constitutes
    It is characterized in that: image display apparatus will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
  5. 5. character recognition system, it is characterized in that: the image input device that has input document image, document image to this input carries out the character recognition treating apparatus that character recognition is handled, the character identification result of this character recognition treating apparatus is presented at display device in the display frame, and the display control unit that when being presented at this character identification result on the display device, shows control, the datum line of display control unit during with the character display recognition result is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character string of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
  6. 6. character recognition system according to claim 5, it is characterized in that: comprise setting device, being used for before the input document, is to erect to write document or write across the page document according to the input document image of importing by operator's operation in advance, sets datum line along laterally or longitudinally showing.
  7. 7. character recognition system according to claim 5 is characterized in that: also be provided with the datum line setting device that the condition of setting datum line is used, display control unit shows datum line according to the condition of the datum line of being set by the datum line setting device.
  8. 8. character recognition system according to claim 7 is characterized in that: the condition of datum line is such, is to erect to write document or write across the page document according to input document image promptly, in advance the datum line edge laterally or is longitudinally shown.
  9. 9. character recognition system according to claim 5, it is characterized in that: when the picture area of character display recognition result image showed as window respectively with the picture area that shows input document image, the lap that datum line is used as both sides' window frame showed.
  10. 10. character recognition system, it is characterized in that: the image input device that has input document image, document image to this input carries out the character recognition treating apparatus that character recognition is handled, the character identification result of this character recognition treating apparatus is presented at display device in the display frame, obtain the accuracy calculation element of accuracy of the character identification result of character recognition treating apparatus, and the display control unit that when being presented at this character identification result on the display device, shows control, the character group of display control unit character display recognition result, when cursor being presented at desire wherein and confirming from the character image of the character of the character identification result here, cross the character of accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
  11. 11. a character identifying method, this method comprises: the stage of input document image a), the document image of input carried out the stage c) of the document image of the image of recognition result of the stage b) of character recognition and character display identification and input,
    It is characterized in that: show in stage c) at this image, will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
  12. 12. the carrier of the software program that a logger computer is used, this software program is used for input document image is carried out character recognition, the image and the input document image of the character identification result that obtains are presented on the picture, and whether the identification of judging character is correct, when the identification of concluding character is wrong, be modified to correct character
    It is characterized in that: will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
  13. 13. the carrier of the software program that a logger computer is used, the stage that this software program is used for carrying out input document image a), the document image of input carried out the stage c) of the document image of the image of recognition result of the stage b) of character recognition and character display identification and input
    It is characterized in that: in image shows stage c), will be adjacent to show from the character of the character identification result that becomes the object that character identification result is judged here and the character that becomes the input document image in character recognition source of character of this character identification result.
  14. 14. carrier, it is characterized in that: this carrier is used for the software program that logger computer is used, the software program that this computing machine is used is used for when the character display recognition result, datum line is presented in the display frame, the character string that shows the character identification result that becomes present character identification result affirmation correction object in a side of this datum line along datum line, showing with a side at datum line shows along datum line at the opposite side of this datum line becomes the character string that present character identification result confirms to revise the corresponding input document image of the character of character identification result of object, and at this moment making becomes present character identification result and confirm that the character string of the character string of character identification result of correction object and the input document image corresponding with it clips datum line mutually and relatively shows.
  15. 15. carrier, it is characterized in that: this carrier is used for the software program that logger computer is used, the software program that this computing machine is used is used in character recognition is handled, calculate the accuracy of character identification result with character identification result, the character group of character identification result is presented in the display frame, when cursor being presented at desire and confirming from the character image of the character of the character identification result here, cross the character of accuracy more than given threshold value, cursor is presented on the character image of accuracy less than the character of given threshold value.
CN 97113677 1996-06-28 1997-06-25 Character recognition result display method, character recognition system and information recording medium Expired - Fee Related CN1103088C (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
JP188341/96 1996-06-28
JP8188341A JPH1021326A (en) 1996-06-28 1996-06-28 Recognized result displaying method, character recognizing system and information recording medium
JP188341/1996 1996-06-28

Publications (2)

Publication Number Publication Date
CN1170912A true CN1170912A (en) 1998-01-21
CN1103088C CN1103088C (en) 2003-03-12

Family

ID=16221928

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 97113677 Expired - Fee Related CN1103088C (en) 1996-06-28 1997-06-25 Character recognition result display method, character recognition system and information recording medium

Country Status (2)

Country Link
JP (1) JPH1021326A (en)
CN (1) CN1103088C (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1333609C (en) * 2002-10-31 2007-08-22 日本电气株式会社 Portable cellular phone provided with character recognition function, method and program for in correctly recognized character
CN100456290C (en) * 2004-11-03 2009-01-28 国际商业机器公司 System and method for automatically and dynamically composing document management application program
CN104680160A (en) * 2013-11-26 2015-06-03 冲电气工业株式会社 Information processing apparatus, system and method
CN108805153A (en) * 2017-05-06 2018-11-13 南京多邦软件有限公司 The concentration of rejected character checks processing method in a kind of intelligence paper identification

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100351584B1 (en) * 2000-07-05 2002-09-05 주식회사 팔만시스템 System of proofreading a Chinese character by contrasting one by one
JP3956114B2 (en) 2002-06-28 2007-08-08 インターナショナル・ビジネス・マシーンズ・コーポレーション Display control method, program using the same, information processing apparatus, and optical character reader
JP2006277001A (en) * 2005-03-28 2006-10-12 Fujitsu Ltd Input image displaying method, and input image displaying program
JP2007305045A (en) * 2006-05-15 2007-11-22 Konica Minolta Medical & Graphic Inc Character reader, id card creation device and id card creation method
JP4873138B2 (en) * 2006-06-21 2012-02-08 富士ゼロックス株式会社 Information processing apparatus and program
JP5316021B2 (en) * 2009-01-26 2013-10-16 富士通株式会社 Clean book support program and clean book support method
JP4919521B2 (en) * 2009-03-30 2012-04-18 孝彰 福田 Character input method and input character proofreading method
JP2012027524A (en) * 2010-07-20 2012-02-09 Sharp Corp Image processor, image processing method and program thereof
JP6464440B1 (en) * 2017-12-27 2019-02-06 株式会社日本デジタル研究所 Accounting processing apparatus, accounting processing system, accounting processing method and program
WO2021238733A1 (en) 2020-05-25 2021-12-02 聚好看科技股份有限公司 Display device and image recognition result display method
CN114339346B (en) * 2020-09-30 2023-06-23 聚好看科技股份有限公司 Display device and image recognition result display method
CN112584213A (en) * 2020-12-11 2021-03-30 海信视像科技股份有限公司 Display device and display method of image recognition result

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1333609C (en) * 2002-10-31 2007-08-22 日本电气株式会社 Portable cellular phone provided with character recognition function, method and program for in correctly recognized character
CN100456290C (en) * 2004-11-03 2009-01-28 国际商业机器公司 System and method for automatically and dynamically composing document management application program
US8112413B2 (en) 2004-11-03 2012-02-07 International Business Machines Corporation System and service for automatically and dynamically composing document management applications
CN104680160A (en) * 2013-11-26 2015-06-03 冲电气工业株式会社 Information processing apparatus, system and method
CN108805153A (en) * 2017-05-06 2018-11-13 南京多邦软件有限公司 The concentration of rejected character checks processing method in a kind of intelligence paper identification

Also Published As

Publication number Publication date
CN1103088C (en) 2003-03-12
JPH1021326A (en) 1998-01-23

Similar Documents

Publication Publication Date Title
CN1103088C (en) Character recognition result display method, character recognition system and information recording medium
US20050216828A1 (en) Patent annotator
US5903666A (en) Methods of splitting and joining handwritten input
CN1763669A (en) Sequence program editing apparatus
US20090309892A1 (en) Information display apparatus, information displaying method, and computer readable medium
JP3814320B2 (en) Image processing method and apparatus
CN103996180A (en) Paper-shredder broken-document restoration method based on English character characteristics
CN1109191A (en) Display unit having plurality of frame buffers
JP2006277001A (en) Input image displaying method, and input image displaying program
JP4661909B2 (en) Information display device and program
CN1383516A (en) Proofreading system of Chinese characters by means of one-to-one comparision
JPH1049623A (en) Character reader
CN1622121A (en) Modified handwritten Chinese character input recognition method
JP4682663B2 (en) Document processing device
JPH0388086A (en) Document reader
CN1097815C (en) Character forming apparatus
CN117076703B (en) Automatic card structured information extraction technical method
CN1928896A (en) Modified hand-written Chinese character input recognition method
CN1084503C (en) Method for automatically correcting truncating error of document and device thereof
JP2504471B2 (en) Text editing device
CN1166152C (en) Data displaying apparatus and data displaying method which can easily discriminate character data when illustration data and character data overlap with each other
JPS63149759A (en) Document editing device
JPH04167085A (en) Handwritten character input device
JPH06251187A (en) Method and device for correcting character recognition error
JPH03240183A (en) Automatic correction system for recognized character

Legal Events

Date Code Title Description
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C06 Publication
PB01 Publication
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20030312

Termination date: 20140625

EXPY Termination of patent right or utility model