JP5316021B2 - Clean book support program and clean book support method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a fair copy support program and a fair copy support method, which can support a fair copy operation of document data appropriately. <P>SOLUTION: The fair copy support program makes a computer run an image display process which displays an image of the document data as a fair copy target, a fair copy character string input process which receives an input of a fair copy character string into the document data, a search process which performs a search procedure with the fair coy character string as a keyword to a character recognition result for the image and obtains position information on the image of the searched character string from the character recognition result, and a display mode change process which changes a display mode of a part relating to the position information obtained in the search process, wherein the display mode change process changes the display mode so that a part relating to characters having a lower conviction degree on the character recognition result than a predetermined value in the part relating to the position information can be discriminated. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a clean-book support program and a clean-book support method, and more particularly, to a clean-book support program and a clean-book support method that support clear-writing work of document data including character strings in a data format other than text.

  In recent years, many tasks are performed by data processing on a computer, but paper is still often used for delivery of forms and the like. For example, when a customer fills in a contract or the like used in a bank or an insurance company, it is almost always written by hand on paper.

  In order to process data on a character string written on a paper document regardless of whether it is a handwritten character or a type, it is necessary to convert the character string written on the paper document into a text character string. As a technique for automatically converting a character string written on a paper document into a text character string, there is an OCR (Optical Character Recognition) technique. In the OCR technique, generally, a line including a character string is cut out from a document image, a delimiter position of characters constituting the line is estimated, and a character image is generated for each character. Further, by recognizing the character image and outputting a character recognition result, a text character string is extracted from the document image. There are various examples of technology for recognizing a character string from a document image by OCR technology (for example, Patent Document 1), and many products are sold.

  However, since the character recognition result has an error, correction work is required. In particular, when the recognition target is a handwritten character or when the image quality of the form image is poor, high recognition accuracy cannot be obtained, and correction work increases. When there are many characters in the document image, it is very difficult for the operator to visually confirm the character recognition result and determine whether or not it is correct, and there is a high possibility of overlooking an error. Therefore, even if the document recognition system is installed, it is not actually used, and there are many cases where the operator inputs everything by hand.

  The operation of inputting a character string written on a paper document into the computer as a text character string is called a clear-writing operation. 2. Description of the Related Art Conventionally, in clean-up work, a paper document is not referred to as it is in order to increase the operator's work efficiency, but the paper document is converted into image data (hereinafter referred to as “document image”) by a scanner or the like, and the document image is converted. The text is input while being displayed on the same screen as the text input area.

  For example, FIG. 1 is a diagram illustrating an example in which a document image and a text input screen are displayed on the same screen. In the example shown in the figure, a document image is displayed on the left side of the screen, and a text input area is displayed on the right side.

  Further, as shown in FIG. 2, in a document image displayed on a screen, an input window may be arranged near a character string to be input to perform text input.

  There are also techniques for reducing the number of characters that an operator must actually input. One example is generally called predictive input, which is installed in many models as a character input function in recent mobile phones. Predictive input predicts a word or a character string that may be connected immediately after a character string input by a keyboard or the like, and presents it as the next input candidate. If a character string that the user wants to input is presented as an input candidate, the user can reduce the burden of the character input operation by selecting the character string.

JP-A-7-262317 Japanese Patent Laid-Open No. 61-07408 JP-A-7-160730

  However, when the operator cleans the text manually, for example, on the screen as shown in FIG. 1, there is a problem that it is difficult to know which part is currently input because the reference position in the document is far from the input position. . On the other hand, when the input position is placed on the document image as shown in FIG. 2, the input window hides the document, so that an operation such as moving the window is required, which is complicated.

  If the input character string is short, the keyword search function can find out where the input character string is in the document. FIG. 3 is a diagram illustrating an example in which a location where an input character string exists is presented by keyword search. In the figure, “Tokyo” is input as a keyword, and an area matching “Tokyo” is searched from two places in the document and highlighted with a dotted line. In such a keyword search, if the document data is converted into text by the OCR technique and then searched, a search error occurs due to a recognition error. Therefore, in Patent Literature 3, a recognition candidate for each character is acquired from a recognition result by the OCR technique, and a keyword is converted into a search expression with reference to the recognition candidate, thereby eliminating a search omission. This is based on the information that “n” in the keyword “computer” is easily mistaken for katakana “ni” or “so” (= “so”), for example, a conditional expression such as “computer or computer or cosopter”. Is to create. However, such a method is impractical because there are a large number of search formulas when the keyword is long, and it can be said that this method is suitable for searching for a short character string. Therefore, there is a problem that practical performance cannot be obtained if the technique of Patent Document 3 is applied as it is to a clearing operation in which a long character string is input.

  Another problem when the operator manually inputs is that it is difficult to notice an input error. Manual input using a keyboard generally has fewer errors than character recognition using OCR technology, but the errors are not completely eliminated. As for the OCR technique, a technique for detecting a portion having a high possibility of recognition error is disclosed (for example, Patent Document 2), but a technique for detecting an error in manual input by an operator is not disclosed. Also, especially in the case of Japanese, since a character string is often input using kana-kanji conversion, the input character string is often partially established as Japanese, and the user is less likely to notice an error. There is also the aspect.

  Further, the manual writing operation requires a large number of man-hours for input, and there are problems such as an operator's work burden and an increase in work costs. Predictive input is regarded as one of the means for mitigating such a problem. However, when the next word is predicted by the general-purpose dictionary, there is a problem in that it is difficult to notice an input error because a character string that is likely to be in Japanese is presented in the same way as input by kana-kanji conversion. On the other hand, when prediction is performed using a character string used by the user in the past, that is, when prediction is performed by the user prediction dictionary, words that are likely to be used in the same document are presented, and thus the possibility of error is reduced. However, it is only effective after inputting a certain amount in the same document. For example, in the document example shown in FIG. 1, words such as “Tokyo” and “foreign” are used several times, so there is a possibility that they can be predicted, but other words appear almost once. Therefore, the input man-hour reduction effect by the user prediction dictionary can hardly be expected.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a clean-up support program and a clean-up support method that can appropriately support the clearing work of document data.

Therefore, in order to solve the above-described problem, the clean-up support program includes an image display procedure for displaying an image of document data as a clear-print target on a display device, and a clear-text string input procedure for receiving a clear-text string input for the document data; A search procedure for executing a search process using the clear character string as a keyword for a character recognition result for the image, and acquiring position information of the searched character string on the image from the character recognition result; and the search procedure And a display mode changing procedure for changing the display mode of the portion related to the position information acquired in step S1, and the display mode changing procedure constitutes the clear text string in the portion related to the position information The display mode is changed so that the part related to the character whose degree of coincidence with the character is lower than a predetermined value can be identified.

  In such a clean-up support program, it is possible to appropriately support clean-up work of document data.

  It is possible to appropriately support the clearing work of document data.

It is a figure which shows the example in which the document image and the text input screen were displayed on the same screen. It is a figure which shows the example by which the window for input was arrange | positioned near the character string of input object. It is a figure which shows the example in which the location where an input character string exists is shown by keyword search. It is a figure for demonstrating the example which improved the visibility of the input position in this Embodiment. It is a figure for demonstrating the method of making an input mistake recognized in this Embodiment. It is a figure for demonstrating the method to reduce the number of input texts in this Embodiment. It is a figure which shows the hardware structural example of the written-book support apparatus in embodiment of this invention. It is a figure which shows the function structural example of the clearing book assistance apparatus in 1st embodiment. It is a flowchart for demonstrating the process sequence by the cleanliness assistance apparatus of 1st embodiment. It is a conceptual diagram for demonstrating intermediate data. It is a figure which shows the structural example of intermediate data. It is a figure which shows the example of conversion from online handwriting data to a document image. It is a figure which shows the example of a division | segmentation of a clear text. It is a figure which shows the example of the search result data by a keyword search part. It is a figure for demonstrating all the combinations which arranged each search result data in order of the corresponding partial character string. It is a figure which shows the example of a pixel scan around the keyword detection area | region for determining a background color. It is a figure which shows the example of the highlight display by changing a background color. It is a flowchart for demonstrating the process sequence of the calculation process of a character string matching degree. It is a figure which shows the example of estimation of the front-end | tip position of the keyword detection area which the search result data after the 2nd show. It is a figure which shows the example of the determination result of whether the vicinity coordinate of an estimated position is contained in the keyword detection area which search result data show. It is a figure for demonstrating the meaning of the co-occurrence relationship determined by the character string matching degree evaluation part. It is a flowchart for demonstrating the process sequence by the cleanliness assistance apparatus of 2nd embodiment. It is a figure which shows the 1st example of the highlight display in 2nd embodiment. It is a figure which shows the 2nd example of the highlight display in 2nd embodiment. It is a flowchart for demonstrating the process sequence by the cleanliness assistance apparatus of 3rd embodiment. It is a figure which shows the function structural example of the cleanliness assistance apparatus in 4th embodiment. It is a flowchart for demonstrating the process sequence by the cleanliness assistance apparatus of 4th embodiment. It is a figure which shows the example of an input object range. It is a figure which shows the function structural example of the cleansing book support apparatus in 5th Embodiment. It is a figure which shows the structural example of a character string estimation part. It is a flowchart for demonstrating the process sequence by the cleanliness assistance apparatus of 5th Embodiment. It is a figure which shows the example by which the background color was changed about the other part except the upper and lower one line of the line to which the character string made into emphasis belongs.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. First, an outline of a solution in the present embodiment for the problem that the input position is difficult to understand will be described.

  FIG. 4 is a diagram for explaining an example in which the visibility of the input position is improved in the present embodiment. As shown in the figure, the clean-up support device of the present embodiment causes the document reference screen 510 and the clean-up screen 520 to be arranged in the horizontal direction and displayed on the display device.

  The document reference screen 510 is a screen for displaying the document data (electronic document) to be printed. The document data is data including a character string in a data format other than a text format (a data format in which characters are recorded by a character code). For example, image data (hereinafter referred to as image data generated by reading a paper document with a scanner) And “document image”))). However, the document data 30 is not necessarily limited to image data. For example, it may be data in which handwriting written by pen input on the screen is stored as a coordinate point sequence (hereinafter referred to as “online handwriting data”).

  On the other hand, the clear text screen 520 is a screen for inputting a text character string corresponding to the document data. In the present embodiment, “clean text” means text conversion of character string information that has not been converted into text (character code). Note that the document reference screen 510 and the clean-up screen 520 need only be displayed on the same display, and do not necessarily have to be adjacent in the horizontal direction. For example, they may be arranged in the vertical direction.

  FIG. 9A shows an example in which “Tokyo,” is input on the clean-up screen 520. At this time, there are two character strings in the document data that are the same as or similar to the input text. The first is “Tokyo,” which is a perfect match, and the second is “Tokyo,” where the reading (,) is missing. The cleanliness support apparatus according to the present embodiment has a location (area) in the document data where there is a character string that is the same as or similar to the input text (that is, a character string that substantially matches (including a complete match) with the input text). To highlight. In the example shown in the figure, the character string in the document data is surrounded by a broken line and underlined by a solid line.

  Subsequently, FIG. 4B shows a state where the input has advanced to “Tokyo, foreign country”. Of the two places highlighted in (A), the character string following the second place does not match “foreign country”. Therefore, the clean-up support apparatus determines that the second part does not correspond to the input text because the range where the second part does not match is long, and excludes the second part from being highlighted. In this way, the portion corresponding to the input text is highlighted, so that the user can easily find the input position in the document data. As a result, the clean-up work can be made efficient.

  Next, an outline of a solution solving method in the present embodiment for the problem that the user is difficult to notice a text input error will be described. In FIG. 4, an area including a character string that is the same as or similar to the input text is highlighted with a dotted line, and an underline is displayed in the character string. The cleanliness support apparatus of the present embodiment displays the underline only on the part that matches the character in the input text. Therefore, when a character that does not match the document data is input due to an input error or the like, the clean-up support apparatus does not display an underline for the character.

  FIG. 5 is a diagram showing a display example for recognizing an input error in the present embodiment. In the figure, an example is shown in which “Saitama Prefecture” is erroneously entered as “Saitama Prefecture” on the screen 520. In this case, the clean-up support apparatus does not display an underline in the part “Saki” on the document reference screen 510. As a result, it can be shown to the user that there is a character that does not partially match in the character string that matches the input text. However, this is only a warning, and the system does not assert that there is an input error. However, since it is a chance to pay attention to a portion where there is a possibility of an input error, it is possible to immediately detect an input error.

  Furthermore, an outline of a method for reducing the number of input texts in the present embodiment will be described. FIG. 6 is a diagram for explaining a method of reducing the number of input texts in the present embodiment.

  The figure shows a state in which “up to Tokyo, foreign country, Saitama prefecture” in the document data is searched as a character string substantially matching the input text. In this state, the clean-up support apparatus according to the present embodiment estimates that the text that is subsequently input is a character string that follows “Saitama Prefecture” because there is only one character string that has been searched. Therefore, the clean-up support device performs character recognition by OCR (Optical Character Recognition) technology for the area following “Saitama Prefecture”, and displays the character string obtained as a recognition result on the clean-screen screen 520 as an input candidate for prediction input. Let As a result, the user only has to select a character string (“transfer”) to be input next from input candidates, and the number of input texts can be reduced. The position of the character searched in the document data in the document image may be determined using the technique described in Patent Document 1. The character size is also known.

  The clean-up support apparatus that realizes the above functions will be described more specifically.

  FIG. 7 is a diagram illustrating a hardware configuration example of the cleanup support apparatus according to the embodiment of the present invention. 7 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, a display device 106, and an input device connected to each other via a bus B. 107.

  A program for realizing the processing in the clean-book support device 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 on which the program is recorded is set in the drive device 100, the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100. However, the program need not be installed from the recording medium 101 and may be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program and also stores necessary files and data.

  The memory device 103 reads the program from the auxiliary storage device 102 and stores it when there is an instruction to start the program. The CPU 104 realizes functions related to the cleanup support apparatus 10 in accordance with a program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network. The display device 106 displays a GUI (Graphical User Interface) or the like by a program. The input device 107 includes a keyboard and a mouse, and is used for inputting various operation instructions.

  FIG. 8 is a diagram illustrating a functional configuration example of the cleanup support apparatus according to the first embodiment. In the drawing, the clean-up support device 10 includes a character recognition unit 111, a document data display unit 112, a clean-text input unit 113, a text division unit 114, a keyword search unit 115, a character string matching degree evaluation part 116, and a character string emphasizing unit. 117 etc. These units are realized by processing that the program causes the CPU 104 to execute.

  The character recognition unit 111 performs character recognition processing using the OCR technique on the document data 30 that is the target of clean-up, and generates character recognition result data 40 in the auxiliary storage device 102 as a character recognition processing result. The document data 30 is data including a character string in a data format other than a text format, such as a document image or online character handwriting data, and is stored in the auxiliary storage device 102, for example.

  The character recognition result data 40 includes not only a confirmed character recognition result (a confirmed character string) but also intermediate data. The intermediate data is data indicating a character recognition result including ambiguity. That is, the intermediate data includes the coordinate information (position information) on the document image and one or more recognition candidates for each recognized character (or each radical of the character), and indicates the certainty factor for each recognition candidate. Includes evaluation value (recognition score) and the like. A character recognition result is determined based on the recognition score in the intermediate data. The character recognition result data 40 includes coordinate information and a recognition score on the document image for each confirmed character string and for each character included in the confirmed character string even in the confirmed character recognition result.

  The document data display unit 112 displays the document reference screen 510 in FIG.

  The clear text input unit 113 displays the clear text screen 520 in FIG. 4 or the like on the display device 106 and accepts input of the clear text 50 (clear text string) via the input device 107. The input clean text 50 input via the input device 107 is recorded in the memory device 103. The clear text input unit 113 displays the clear text 50 recorded in the memory device 103 on the clear text screen 520. In the present embodiment, the clear text means character string data in a text format.

  The text dividing unit 114 divides the clear text 50 into partial character strings having an appropriate length.

  The keyword search unit 115 performs a keyword search for the character recognition result data 40 for each divided partial character string. By the keyword search, the position (coordinate region) where the partial character string exists in the document data 30 and the evaluation value (search score) indicating the degree of coincidence or certainty are obtained. The keyword search method is not limited to a specific one. For example, a known search method such as the method described in Patent Document 3 may be employed.

  The character string matching degree evaluation unit 115 calculates the character string matching degree. The character string matching degree refers to an evaluation value indicating a high possibility that a character string searched for each partial character string from the character recognition result data 40 is continuous on the document image.

  The character string emphasizing unit 117 selects a partial character string having a character string matching degree higher than a predetermined threshold (predetermined value) from the partial character strings searched from the document data 30 by the keyword searching unit 115. The character string emphasizing unit 117 highlights a portion (area) where the selected partial character string exists in the document data 30.

  Hereinafter, the processing procedure of the clean-book support apparatus 10 of FIG. 8 is demonstrated. FIG. 9 is a flowchart for explaining a processing procedure by the cleansing support apparatus according to the first embodiment.

  In step S <b> 110, the character recognition unit 111 acquires the document data 30 that is the target of clean-up from the auxiliary storage device 102, and executes character recognition processing using the OCR technique on the entire document data 30. The character recognition unit 111 generates character recognition result data 40 including intermediate data in the auxiliary storage device 102 as a result of the character recognition process.

  FIG. 10 is a conceptual diagram for explaining the intermediate data. The intermediate data is data generally called a lattice. In the figure, (A) shows a handwritten character string to be recognized. Also, (B) shows a conceptual diagram of the lattice. As shown in (B), in the lattice, the character string to be recognized is divided into small groups, and transition links are provided between the groups. For example, a link (arrow) extending from “tree” to “na” means that “tree” and “卆” between the links become one character. In this case, the character recognition result is “frame”.

  FIG. 11 is a diagram illustrating a configuration example of intermediate data. In the figure, a specific example of intermediate data of the character string shown in FIG. 10 is shown. As shown in the figure, in the lattice, one data term is assigned to each unit obtained by finely grouping character strings to be recognized. Each data term includes a character or a part of the character corresponding to the data term, a coordinate value, and the like. Each data item includes a link destination, a character recognition result, a recognition score, and the like for each link destination (for each link 1, link 2, link 3, and the like). For example, in the data item of item number 1 in FIG. 11, the link destination of link 1 is 2, the character recognition result is “tree”, and the recognition score is 250. A plurality of character recognition results (recognition candidates) can exist for the same character. A plurality of recognition candidates for the same character can be judged as superior or inferior by the recognition score for each. A known technique may be used as a method for generating intermediate data (lattice). In carrying out this embodiment, the intermediate data does not necessarily have the configuration shown in FIG. For example, a simple configuration in which a recognition candidate is included for each recognized character may be used.

  In step S110, information regarding the confirmed character string confirmed based on the intermediate data as shown in FIG. Note that step S110 may be executed continuously (synchronously) with step S120 and subsequent steps. It may be executed in advance asynchronously with step S120 and subsequent steps. Alternatively, it may be executed on another computer different from the cleansing support apparatus 10.

  When the document data 30 is a coordinate point sequence such as online handwriting data, for example, the bitmap coordinates between the coordinate points are converted into a document image by a process of filling in black pixels, and a character is applied to the document image. The recognition unit 111 may perform the character recognition process.

  FIG. 12 is a diagram illustrating an example of conversion from online handwriting data to a document image. In the drawing, an example is shown in which the line segment indicated by the two coordinate points in (A) is converted into a document image (raster data) as shown in (B). Note that details of the conversion process are well-known techniques, and a description thereof will be omitted.

  Subsequently, the document data display unit 112 displays the document reference screen 510 on the display device 106, and displays the document image of the document data 30 that is the target of the clean-up in the document reference screen 510. Further, the clear text input unit 113 displays the clear text screen 520 side by side in the horizontal direction of the document reference screen 510 (S120).

  When the clear text work is started and the clear text 50 is sequentially input on the clear text screen 520 by the user, the clear text input unit 113 determines whether the text being input is in a fixed state every time one character is input (S130). Is determined (S140). Here, the finalized state means a state where the kana-kanji conversion is completed, for example. When the text being input is in a finalized state, the clear text input unit 113 adds the text being input to the clear text 50.

  Subsequently, the text dividing unit 114 divides all the clean text 50 input so far into partial character strings according to a predetermined division rule (S150). Various division rules can be adopted. For example, the character may be divided every fixed number of characters (for example, three characters) from the last input character. Or you may divide | segment immediately after a punctuation mark. Or you may divide | segment for every division | segmentation of character types, such as Kanji, Kana, or an alphanumeric character. Or you may divide | segment by a morphological analysis technique using a word dictionary. Or, in the case of a number, it may be divided by a comma expressing a digit. Further, if the length of the divided partial character string is 8 or more, the divided partial character strings may be equally divided so that all the divided partial character strings have a length of 4 or more. These division rules are only examples.

  FIG. 13 is a diagram illustrating an example of division of clean text. The figure shows a state where the clear text 50 (“Tokyo, foreign country, Saitama prefecture”) is divided for each punctuation mark. Therefore, it is divided into three partial character strings of “Tokyo,” “Foreign,” and “Saitama Prefecture”.

  Subsequently, the keyword search unit 115 performs a keyword search using the partial character string as a keyword for each of the divided partial character strings on the character recognition result data 40, and characters that are the same as or similar to the partial character string in the document image The position where the column exists (hereinafter referred to as “keyword detection area”) is specified (S160). In the keyword search, for example, a technique described in Patent Document 3 may be used. That is, a similar keyword for each keyword is generated based on the intermediate data, and a search for the confirmed character string is performed using a search expression in which the keyword and the similar keyword are ORed. According to such a keyword search, an ambiguous search can be executed. That is, the keyword search unit 115 can specify not only the position of the character string that completely matches the partial character string but also the position of a similar character string. However, an ambiguous search may be performed using other known techniques. Moreover, you may perform a keyword search, without giving ambiguity to a keyword. In this case, an existing position (keyword detection area) of the same character string as the keyword is specified in the document image.

  The keyword search unit 115 records the search result as search result data in the memory device 103. FIG. 14 is a diagram illustrating an example of search result data by the keyword search unit. In the figure, an example of search result data for a partial character string (keyword) “Tokyo” is shown.

  As shown in the figure, the search result data includes information such as a keyword, a search score, a character score, character string coordinates, and character coordinates. The search score is an evaluation value indicating the degree of coincidence between the keyword and the character string searched from the intermediate data, or the certainty of the search result. In the present embodiment, the total value of the character scores is set as the search score. The character score is an evaluation value indicating the degree of coincidence or certainty for each character constituting the keyword, and is generated based on the certainty factor included in the character recognition result data 40.

  The character string coordinates are coordinate values (coordinate values on the document image) of two vertices on a diagonal line of a rectangular area (that is, a keyword detection area) surrounding the entire searched character string on the document image. The character coordinates are the coordinate values of two vertices on the diagonal line of the rectangular area surrounding the character for each character constituting the searched character string. In FIG. 14, “,” of “Tokyo,” is not included, but this is for convenience.

  The search result data is generated for each partial character string (or for each keyword when a plurality of keywords are generated based on one partial character string). As shown in FIG. 4A, when a plurality of search results are obtained for one partial character string, search result data is generated for each search result. Therefore, a plurality of search result data can be generated for one partial character string.

  Subsequently, the character string match evaluation unit 116 evaluates the possibility that the text detection areas are consecutively located for all combinations in which the search result data are arranged in the order of the corresponding partial character strings. The value (character string matching degree) is calculated (S170). Here, all the combinations in which the search result data are arranged in the order of the corresponding partial character strings refer to the combinations shown in FIG.

  In FIG. 15, two search result data are output for the partial character string “Tokyo,”, two search result data are output for the partial character string “foreign,” and one search result data is output for “Saitama Prefecture”. An output example is shown. When these search result data are arranged in the order of partial character strings corresponding to each search result data, there are four combinations of search result data. These four patterns are “all combinations in which each search result data is arranged in the order of corresponding partial character strings” in FIG. Note that the calculated character string matching degree is given for each combination and recorded in the memory device 103.

  Subsequently, the character string emphasizing unit 117 performs highlighting based on each search result data included in a combination in which the character string matching degree given by the character string matching degree evaluating unit 115 is larger than a predetermined threshold value (reference value). The position to perform is determined (S180). Specifically, the character string emphasizing unit 117 sets the minimum rectangular area including the keyword detection areas (three keyword detection areas if the number of partial character strings is 3) of all search result data related to the combination. Based on the character string coordinates of the search result data, the minimum rectangular area is determined as a position (area) for highlighting.

  Subsequently, the character string emphasizing unit 117 highlights the minimum rectangular area determined in step S180 (S190). Specifically, the character string emphasizing unit 117 draws a broken line around the minimum rectangular area. Further, the character string emphasizing unit 117 draws a solid line on the bottom of the rectangular area. However, only one of the broken line and the solid line may be drawn.

  Steps S130 to S190 are repeated until the clearing work is completed (S210).

  By the way, the highlighting method is not limited to the method of surrounding with a broken line or underlining. For example, the background color of the character string to be highlighted may be changed. In this case, the background color may be a predetermined color or may be automatically determined according to the state of the document image. When the background color is automatically determined, first, the original (current) background color of the corresponding area is obtained. Specifically, the corresponding area is binarized, and each pixel is converted into one of monochrome pixels. As a binarization method, for example, “Nobuyuki Otsu,“ Automatic threshold selection method based on discriminant and least squares criterion ”, Electronic Notification Society paper vol. 4, April 1980, p349-365 ”may be used. Subsequently, for example, as shown in FIG. 16, the pixels are scanned in order along the periphery (circumscribed rectangle) of the keyword detection area, and the values of the pixels outside the keyword detection area are obtained in order. In the binarized image, the number of pixels and the average value of the pixels that are white pixels and the pixels that are black pixels are obtained, and the value with the larger number of pixels is determined as the original background color.

  When emphasizing the keyword detection region, the background pixel (background pixel) is specified with reference to the previously obtained binary image, and the emphasized color for the original background color is obtained. Subsequently, the pixel value of the background pixel is replaced with the pixel value of the emphasized color. For example, when the background color is white as shown in FIG. 16 and the highlight color is gray, the result is as shown in FIG.

  Next, details of step S170 will be described. FIG. 18 is a flowchart for explaining the processing procedure of the character string matching degree calculation processing.

  In step S301, the character string matching degree evaluation unit 116 determines whether or not there are a plurality of partial character strings. For example, the state of FIG. 4A is a case where there is one partial character string. The state of FIG. 4B is a case where there are a plurality (two) of partial character strings.

  When there are a plurality of partial character strings (Yes in S301), the character string matching degree evaluation unit 116 estimates the character size of the character string included in the keyword detection area for each search result data of the first partial character string ( S302). In the example of FIG. 15, the character size is estimated for the search result data 1-1 and the search result data 1-2. The character size may be estimated as follows, for example. First, the height and width are obtained for each character based on the character coordinates included in the search result data, and the maximum value and the average value are calculated. Subsequently, an intermediate value between the maximum value and the average value is obtained, and the intermediate value is set as an estimated value of the character size. When estimating the character size, it is preferable to exclude characters with special sizes such as punctuation marks from the calculation target. By doing so, the character size can be estimated stably.

  Subsequently, the character string matching degree evaluation unit 116 calculates the difference (that is, the difference in the number of characters) on the character code from the end of the first partial character string to the front end of the other partial character string in the arrangement order before the division. Obtain (S303). For example, in the case of FIG. 13, the first partial character string in the arrangement order before division is “Tokyo,” and the other partial character strings are “foreign,” and “Saitama Prefecture”. In this case, the numbers of characters from the end of “Tokyo,” to the tips of “foreign,” and “Saitama” are 0 and 3, respectively.

  Subsequently, the character string matching degree evaluation unit 116 estimates the tip position (coordinate value on the document image) of the keyword detection region corresponding to the second and subsequent partial character strings by multiplying the difference in the number of characters by the character size. And assigned to each search result data (S304). For example, assuming that the character size estimated based on the search result data 1-1 of “Tokyo,” in the example of FIG. 15 is 30 dots, 0 × 30 = 0 dot movement from the end of “Tokyo,” It is estimated that the moved position is the tip of “Foreign”, and the position moved by 3 × 30 = 90 dots is the tip of “Saitama Prefecture”. The end of “Tokyo,” may be the end of the rectangular area (that is, the keyword detection area) indicated by the character string coordinates of the search result data of “Tokyo,”.

  Note that the tip position of the keyword detection area corresponding to the second and subsequent partial character strings is estimated for each of the first partial character strings. Specifically, in the example of FIG. 15, as shown in FIG. 19, the tip position of the keyword detection region indicated by the second and subsequent search result data is estimated.

  FIG. 19 is a diagram illustrating an example of estimating the tip position of the keyword detection area indicated by the second and subsequent search result data.

  In the figure, for the search data 2-1 and 2-2, there are an estimated position (x21, y21) based on the search result data 1-1 and an estimated position (x22, y22) based on the search result data 1-2. Has been granted. Further, an estimated position (x31, y31) based on the search result data 1-1 and an estimated position (x32, y32) based on the search result data 1-2 are given to the search result data 3-1.

  Subsequently, the character string matching degree evaluation unit 116 determines whether or not there is a co-occurrence relationship between each search result data of the first partial character string and each search result data of the second and subsequent partial character strings (S305). . In the present embodiment, when the vicinity coordinates of the estimated position are included in the keyword detection area indicated by the search result data, the search result data of the first partial character string used as the base point of the estimated position and the keyword detection area It is determined that the corresponding search result data has a co-occurrence relationship.

  The vicinity coordinates of the estimated position means (x ± α, y ± α) (α is a character size, for example) when the coordinate value of the estimated position is (x, y). Accordingly, in FIG. 19, for the search result data 2-1, the search result data 2-1 indicates the coordinate value of (x21 ± α, y21 ± α) or the coordinate value of (x22 ± α, y22 ± α). It is determined whether or not it is included in the keyword detection area. The same determination is performed for the search result data 2-2 and the search result data 3-1.

  FIG. 20 is a diagram illustrating an example of a determination result of whether or not the vicinity coordinates of the estimated position are included in the keyword detection region indicated by the search result data. In the same figure, the estimated position marked with “x” indicates that it is not included in the keyword detection area indicated by the corresponding search result data. The estimated position marked with ○ is included in the keyword detection area indicated by the corresponding search result data.

  In the figure, it is shown that any estimated position is not included in the keyword detection area indicated by the search result data 2-1. Moreover, it is shown that the keyword detection area indicated by the search result data 2-2 includes only the estimated position (x21, y21) based on the search result data 1-1. Moreover, it is shown that the keyword detection area indicated by the search result data 3-1 includes only the estimated position (x31, y31) based on the search result data 1-1.

  Therefore, it is determined that the search result data 1-1 and the search result data 2-2, and the search result data 1-1 and the search result data 3-1 have a co-occurrence relationship. That is, the keyword detection area indicated by the search result data 1-1, 2-2, and 3-1 searched for each partial character string is determined to be a continuous area as shown in FIG. it can.

  Note that the continuity of each keyword detection area may be determined by directly comparing and collating the character string coordinates of each search result data. In this case, steps S302 to S304 may not be executed.

  Subsequently, the character string matching degree evaluation unit 116 calculates the character string matching degree based on the presence or absence of the co-occurrence relationship with the first search result data for every combination of the search result data in the partial character string order. (S306). The calculation of the character string matching degree for each combination is performed as follows, for example. First, the search score of the first search result data of the combination is set as the initial value of the evaluation value. Subsequently, a search score or 0 is added to the evaluation value based on whether or not the second and subsequent search result data of the combination has a co-occurrence relationship with the first search result data. That is, only the search score of the search result data having a co-occurrence relationship with the first search result data is added to the evaluation value. Subsequently, the evaluation value is divided by the number of search result data belonging to the combination (3 in the example of FIG. 20). The result of this division is the string matching degree.

  In the example of FIG. 20, the character string matching degree is calculated for each of the four combinations. At this time, the search score of the search result data 2-1 is not an addition target.

  On the other hand, when there is one partial character string, the character string matching degree evaluation unit 116 calculates the search score of one or a plurality of search result data for the partial character string as the character string matching degree of each search result data. To each search result data (S307). When step S307 is executed, in step S180 of FIG. 9, a broken line and an underline are drawn for the keyword detection area indicated by the search result data having a character string matching degree equal to or higher than a predetermined threshold.

  As described above, according to the first embodiment, a portion that has been clarified in the document data 30 is highlighted. Therefore, for example, it is possible to reduce an input error such as an input being shifted by one line from the middle. As a result, it is possible to reduce the work burden on the user in the clearing work.

  In the present embodiment, the clean text 50 is divided into partial character strings, and a keyword search is performed for each partial character string. Therefore, it is possible to improve the processing efficiency of the keyword search related to the clear text 50 that is likely to be a long sentence. For example, consider a case where the method described in Patent Document 3 is used as a keyword search. For the convenience of explanation, it is assumed that the clear text 50 is divided into three partial character strings, and four keywords can be created for each. In this case, assuming that the character string is not divided into partial character strings, a keyword search is performed for 4 × 4 × 4 keywords with respect to the clear character string 50. On the other hand, if it is divided into partial character strings, a keyword search may be performed for 4 × 3 keywords. Therefore, the processing efficiency of search processing can be significantly improved.

  However, when a keyword search is performed for each partial character string, there is a possibility that a possibility that a portion that does not correspond to the clear text position is determined as a keyword detection region is increased. Therefore, in the present embodiment, the continuity of each keyword detection area is evaluated by the character string matching degree evaluation unit 115, and whether or not each keyword detection area is a location corresponding to the input position based on the evaluation result. Is determined. Thereby, the said side effect is suppressed appropriately.

  In FIG. 9, the example in which the process after step S150 is executed at the timing when the input text is in the finalized state has been described, but other timings may be used as a trigger for executing the process after step S150. For example, the timing when a space is input may be used, or step S150 may be executed at regular intervals.

  Next, a second embodiment will be described. In the second embodiment, differences from the first embodiment will be described.

  FIG. 22 is a flowchart for explaining a processing procedure by the cleansing support apparatus according to the second embodiment. In FIG. 22, the same steps as those in FIG. 9 are denoted by the same step numbers, and description thereof will be omitted as appropriate.

  In FIG. 22, step S185 is added. In step S185, the character string emphasizing unit 117 selects a character from among the search result data included in the combination in which the character string matching degree given by the character string matching degree evaluation unit 115 is larger than a predetermined threshold (reference value). Character coordinates of characters whose score is lower than a predetermined threshold are acquired.

  When highlighting is performed in the subsequent step S190, the character string emphasizing unit 117 changes the display mode so that the character coordinate area acquired in step S185 can be distinguished from other highlighted parts (emphasis). Change display attributes or exclude them from highlighting). In the present embodiment, the area is excluded from the underline drawing target.

  FIG. 23 is a diagram illustrating a first example of highlighting in the second embodiment. The figure shows an example in which “Saitama” in “Saitama Prefecture” is erroneously entered in “Aya”. In this case, in the search result data for “Saitama Prefecture”, the character score of “Sai” is significantly low. Therefore, in step S185, it is determined that the character score of “Aya” is lower than a predetermined threshold. As a result, the “color” area is excluded from the underline drawing target.

  In addition, you may make it change the display mode of the part which does not correspond partially by another method. FIG. 24 is a diagram illustrating a second example of highlighting in the second embodiment. In the figure, an example is shown in which the background color is changed except for the “Saki” area where an erroneous character is input.

  As described above, according to the second embodiment, it is possible to easily show to the user that there are characters that do not partially match in the highlighted portion.

  Next, a third embodiment will be described. In the third embodiment, differences from the first embodiment will be described. FIG. 25 is a flowchart for explaining a processing procedure by the cleansing support apparatus according to the third embodiment. In FIG. 25, the same steps as those in FIG. 9 are denoted by the same step numbers, and description thereof will be omitted as appropriate.

  In FIG. 25, step S150 is replaced with step S155. In step S155, the text dividing unit 114 divides a part from the end of the clean text 50 into partial character strings. That is, in the third embodiment, not a part of all the clean text 50 input so far but only a part from the end is targeted for division. The part to be divided is, for example, immediately after the first punctuation mark found at a position exceeding a predetermined number of characters (for example, 20 characters) within the range of the maximum number of characters (for example, 40 characters). The range is up to the last character. Alternatively, only the text newly determined to be in the finalized state in step S140 may be targeted for division. Furthermore, the division target portion may be determined by other methods.

  Since only the last part of the clean text 50 is to be divided, the number of partial character strings to be processed in step S160 and later can be reduced. Therefore, when the clear text 50 is very long, the calculation cost can be reduced. In addition, when the middle of the input text is edited, only the vicinity of the edited position can be highlighted, so that the convenience can be improved.

  Next, a fourth embodiment will be described. FIG. 26 is a diagram illustrating a functional configuration example of the cleansing support apparatus according to the fourth embodiment. In FIG. 26, the same parts as those in FIG.

  In the figure, the clean-up support device 10a includes an input position determination unit 118, an input target range estimation unit 119, an input target range character recognition unit 120, an estimated character string candidate generation unit 121, a predicted character string display unit 122, and a predicted character string. It further has a selection unit 123 and the like.

  The input position determination unit 118 obtains the current input position (clearance position). For example, in the example of FIG. 6, the coordinate value at the end of the “Saitama Prefecture” area is determined as the input position. Note that the position of the end of “Saitama Prefecture” may be determined based on the character string coordinates of the search result data of “Saitama Prefecture”.

  Based on the input position determined by the input position determination unit 118, the input target range estimation unit 119 estimates a region (hereinafter referred to as “input target range”) that includes a character string to be printed next in the document image. . For example, in the example of FIG. 6, an area including “transfer” following “Saitama Prefecture” is determined as the input target area.

  The input target range character recognition unit 120 executes character recognition processing using the OCR technique on the input target range. For example, in the example of FIG. 6, character recognition is performed on an area including “transfer”. Note that the processing result of the character recognition processing may include a plurality of recognition candidates instead of the character recognition confirmation result, similarly to the processing result by the character recognition unit 111.

  Based on the recognition candidates output from the input target range character recognition unit 120, the estimated character string candidate generation unit 121 is a list of character string candidates estimated to be input following the input position (hereinafter, “estimated character string”). "Candidate list").

  The predicted character string display unit 122 displays a pop-up window 530 for selecting an input text from the estimated character string candidate list on the clean-up screen 520 as shown in FIG.

  The predicted character string selection unit 123 adds the character string selected in the pop-up window 530 to the clean text 50 as a new input text.

  Hereinafter, a processing procedure of the clean-book support apparatus 10a of FIG. FIG. 27 is a flowchart for explaining a processing procedure performed by the cleansing support apparatus according to the fourth embodiment. In FIG. 27, the same steps as those in FIG. 9 are denoted by the same step numbers, and description thereof will be omitted as appropriate.

  In FIG. 27, steps S191 to S202 are added. In step S191, the input position determination unit 118 determines the current input position. Specifically, the input position determination unit 118 includes an area in the document image in which the character string matching degree calculated by the character string matching degree evaluation unit 117 is higher than a predetermined threshold (that is, an area to be highlighted). ) Is determined as the current input position.

  Subsequently, the input target range estimation unit 119 estimates the input target range based on the input position determined by the input position determination unit 118 (S192). For example, in the case of horizontal writing, the input target range estimation unit 119 specifies a coordinate value (for example, coordinates of two vertices on a diagonal line of the rectangular area) that specifies a rectangular area of a predetermined number of characters in the right direction from the input position. Value). The character size for one character in the keyword detection area immediately before the input position is estimated by the same method as the character size estimation method by the character string matching degree evaluation unit 115 described above, and a predetermined number of characters are determined based on the character size. The rectangular area may be determined. For example, an area for N characters may be obtained by finding an area having a width N times the character size to the right from the input position and a height twice the character size in the vertical direction.

  FIG. 28 is a diagram illustrating an example of the input target range. In the figure, a rectangular area surrounding “Saitama Prefecture” is the immediately preceding keyword detection area. A dashed rectangular area adjacent to the right of “Saitama Prefecture” is an input target range. The input target range is set so that the center in the height direction matches the center in the height direction of the immediately preceding keyword detection region. An arrow in the input target range indicates the width of the character size. Accordingly, this figure corresponds to an example in which three characters are set as the input target range.

  Note that a rectangular area having a fixed size may be set as the input target range.

  Whether the character string in the document image is written vertically or horizontally may be determined based on the character coordinates of the search result data based on whether the characters are arranged in the vertical or horizontal direction.

  The input target range character recognition unit 120 performs a character recognition process on the input target range and outputs one or more recognition candidates (S193). Note that the input target range character recognition unit 120 may not perform character recognition again. For example, recognition candidates for the input target range may be output based on the character recognition result data 40.

  Subsequently, the estimated character string candidate generation unit 121 generates a list including recognition candidates output from the input target range character recognition unit 120 in the memory device 103 as an estimated character string candidate list (S194).

  Subsequently, the predicted character string display unit 122 displays a pop-up window 530 for selecting an input text from the estimated character string candidate list on the clean-up screen 520 as shown in FIG. 6 (S201).

  When an input text is selected from the estimated character string candidate list in the pop-up window 530 (Yes in S202), the predicted character string selection unit 123 adds the selected character string to the clean text 50 as a new input text. (S197). Subsequent to step S197, step S150 and subsequent steps are repeatedly executed.

  On the other hand, if the input text is not selected from the estimated character string candidate list in the pop-up window 530 (No in S202), the process proceeds to step S210.

  As described above, according to the fourth embodiment, clear text input candidates for the input target range are presented on the pop-up window 530 based on the character recognition result for the input target range. Therefore, when a correct character string exists in the presented input candidates, the character string can be input only by performing a selection operation. As a result, it is possible to reduce the work load and increase the efficiency of the clearing work. In particular, when the accuracy of character recognition is high, a clearing operation can be performed by a continuous selection operation.

  In addition, you may combine 2nd or 3rd embodiment with 4th embodiment. That is, steps S185 and S155 may be executed in FIG. In particular, when combined with the second embodiment, even if an erroneous candidate (a candidate that is highly likely to be incorrect) is selected in the pop-up window 530, the user is notified of the erroneous part (character). Can do.

  Next, a fifth embodiment will be described. FIG. 29 is a diagram illustrating an example of a functional configuration of the cleanup support apparatus according to the fifth embodiment. In FIG. 29, the same portions as those in FIG. 26 are denoted by the same reference numerals, and the description thereof is omitted.

  In the figure, the clean-up support apparatus 10b further includes a character string predicting unit 124 and a predicted character string integrating unit 125.

  The character string predicting unit 124 uses a known predictive input technique to list a character string candidate that is likely to be input following the character string (word) at the end of the clean text 50 (hereinafter, “predicted character string”). "Candidate list").

  The predicted character string integration unit 125 integrates the predicted character string candidate list generated by the character string prediction unit 124 and the estimated character string candidate list generated by the estimated character string candidate generation unit 121.

  FIG. 30 is a diagram illustrating a configuration example of the character string prediction unit. In the figure, the character string prediction unit 124 includes a word input unit 1241, a word learning unit 1242, a general dictionary 1243, a user prediction dictionary 1244, and the like.

  The word predicting unit 1241 determines a character string (predicted character) that may be subsequently input based on the already input character string (clear text 50 in the present embodiment), the general dictionary 1243, and the user prediction dictionary 1244. Column). The word learning unit 1242 is a character string whose input has been confirmed by the user (in this embodiment, a character string that the prediction character string selection unit 123 has accepted selection, or a character string that the clear text input unit 113 has accepted input confirmation. ) In the user prediction dictionary 1244 in association with the immediately preceding character string. In the general-purpose dictionary 1243, words that are likely to be used together are stored together with a mutual connection relationship. The character string prediction function including the general dictionary 1243 and the user prediction dictionary 1244 is described in “Toshiyuki Masui,“ Fast Text Input Method Using Pen ”. Modern Science, Interactive Systems and Software IV: Japan Software Science Society WISS'96. pp. 51-60, December 1996 "shows an example.

  Hereinafter, the processing procedure of the clean-book support apparatus 10b of FIG. FIG. 31 is a flowchart for explaining a processing procedure by the cleansing support apparatus according to the fifth embodiment. In FIG. 31, the same steps as those in FIG. 27 are denoted by the same step numbers, and description thereof will be omitted as appropriate.

  In FIG. 31, steps S195 and S196 are added. In step S195, the character string predicting unit 124 uses a known predictive input technique to list character string candidates (predicted characters) that are likely to be input following the last character string (word) of the clear text 50. Column candidate list) is generated in the memory device 103. Subsequently, the predicted character string integration unit 125 integrates the estimated character string candidate list and the predicted character string candidate list generated in the memory device 103, and records the integration result in the memory device 103 (S196).

Here, the predicted character string candidate list generated by the character string predicting unit 124 is a list of character strings obtained from the text input so far, and is effective if the same character string is not used many times. Cannot predict. Therefore, it is considered that the estimated character string candidate list generated by the estimated character string candidate generation unit 121 is more likely to match the subsequently input character string. Therefore, in the integration result, it is preferable to arrange the character strings included in the estimated character string candidate list at the top and then arrange the character strings included in the predicted character string candidate list. However, the character strings included in the estimated character string candidate list that have a recognition score (character recognition certainty) lower than a predetermined value may be arranged after the character strings included in the predicted character string candidate list. By doing so, a character string with high reliability can be preferentially presented as an input candidate. In subsequent step S201, the predicted character string display unit 122 causes the pop-up window 530 to display the integrated character string candidate list. .

  As described above, according to the fifth embodiment, a character string predicted by the prediction input function is also presented as an input candidate in addition to the character string that has been character-recognized. Therefore, even if the accuracy of character recognition is low, an appropriate input candidate can be presented. As a result, it is possible to reduce the work load and increase the efficiency of the clearing work.

  By the way, in the first to fifth embodiments, the character string emphasizing unit 117 may change the background color for other portions except the upper and lower lines of the line to which the character string to be emphasized belongs. .

  FIG. 32 is a diagram illustrating an example in which the background color is changed for the other parts except one line above and below the line to which the character string to be emphasized belongs.

  This figure shows a display example of the document reference screen 510 when “Tokyo, foreign country, Saitama prefecture” is input and only “Sai” is mistyped as shown in FIG. In FIG. 5, the background color is changed as a highlighting method, and the correct part is underlined.

  Furthermore, the background color is also gray for the outer portion of the row to which “Tokyo, Foreign, Saitama Prefecture” belongs, except for one row above and below. In this way, by shadowing an area slightly away from the location that is currently being printed, the area that is currently being written can be displayed with more emphasis.

  As mentioned above, although the Example of this invention was explained in full detail, this invention is not limited to such specific embodiment, In the range of the summary of this invention described in the claim, various deformation | transformation・ Change is possible.

Regarding the above description, the following items are further disclosed.
(Appendix 1)
An image display procedure for causing the display device to display an image of the document data that is the target of the copy,
A clear character string input procedure for receiving input of a clear character string for the document data, and a search process using the clear character string as a keyword with respect to a character recognition result for the image, and the searched character string on the image A search procedure for acquiring position information from the character recognition result;
Causing the computer to execute a display mode change procedure for changing the display mode of the portion related to the position information acquired in the search procedure;
The display mode changing procedure is a clean-up support program for changing a display mode so that a portion related to a character whose certainty of the character recognition result is lower than a predetermined value in the portion related to the position information can be identified.
(Appendix 2)
A division procedure for dividing the clear character string into a plurality of partial character strings;
The cleanup support program according to supplementary note 1, wherein the search procedure executes the search process for each partial character string and acquires the position information of the character string searched for each partial character string.
(Appendix 3)
A determination procedure for determining whether the character string searched for each partial character string is a continuous character string on the image based on the position information of the character string searched for each partial character string; ,
The cleanup support program according to supplementary note 2, wherein the display mode changing procedure changes the display mode when it is determined that the character string searched for each partial character string is a continuous character string on the image.
(Appendix 4)
The cleanup support program according to any one of appendices 1 to 3, wherein the search procedure executes the search process using a part from the end of the clear text string.
(Appendix 5)
Additional remark 1 to 4 which has the input candidate display procedure which displays the character string obtained by the character recognition with respect to the predetermined range from the end of the part concerning the position information acquired in the search procedure as the next input candidate Clean book support program.
(Appendix 6)
The cleanup support program according to supplementary note 5, wherein the input candidate display procedure displays a character string predicted to be input following the character string at the end of the clear character string as a next input candidate.
(Appendix 7)
A clean-up support method executed by a computer,
An image display procedure for causing the display device to display an image of the document data that is the target of the copy,
A clear character string input procedure for receiving input of a clear character string for the document data, and a search process using the clear character string as a keyword with respect to a character recognition result for the image, and the searched character string on the image A search procedure for acquiring position information from the character recognition result;
A display mode change procedure for changing the display mode of the portion related to the position information acquired in the search procedure,
The said display mode change procedure is a cleanliness support method which changes a display mode so that the part which concerns on the character whose reliability of the said character recognition result is lower than a predetermined value among the parts which concern on the said position information can be identified.
(Appendix 8)
A division procedure for dividing the clear character string into a plurality of partial character strings;
The cleanup support method according to appendix 7, wherein the search procedure executes the search process for each partial character string, and acquires the position information of the character string searched for each partial character string.
(Appendix 9)
A determination procedure for determining whether the character string searched for each partial character string is a continuous character string on the image based on the position information of the character string searched for each partial character string; ,
9. The cleanliness support method according to appendix 8, wherein the display mode change procedure changes the display mode when it is determined that the character string searched for each partial character string is a continuous character string on the image.
(Appendix 10)
The cleanup support method according to any one of appendices 7 to 9, wherein the search procedure executes the search process using a part from the end of the clear text string.
(Appendix 11)
The supplementary note 7 to any one of claims 7 to 10, further comprising an input candidate display procedure for displaying a character string obtained by character recognition for a predetermined range from the end of the portion related to the position information acquired in the search procedure as a next input candidate. Clean-book support method.
(Appendix 12)
12. The cleanup support method according to supplementary note 11, wherein the input candidate display procedure displays a character string predicted to be input following the character string at the end of the clear character string as a next input candidate.

DESCRIPTION OF SYMBOLS 10 Formatting support apparatus 100 Drive apparatus 101 Recording medium 102 Auxiliary storage apparatus 103 Memory apparatus 104 CPU
105 interface device 106 display device 107 input device 111 character recognition unit 112 document data display unit 113 clear text input unit 114 text segmentation unit 115 keyword search unit 116 character string matching degree evaluation 117 character string emphasizing unit 118 input position determination unit 119 input Target range estimation unit 120 Input target range character recognition unit 121 Estimated character string candidate generation unit 122 Predicted character string display unit 123 Predicted character string selection unit 124 Character string prediction unit 125 Predicted character string integration unit 1241 Word input unit 1242 Word learning unit 1243 General-purpose dictionary 1244 User prediction dictionary B bus

Claims (6)

An image display procedure for causing the display device to display an image of the document data that is the target of the copy,
A clear character string input procedure for accepting input of a clear character string for the document data ;
A search procedure for executing a search process using the clear character string as a keyword for the character recognition result for the image, and obtaining position information on the image of the searched character string from the character recognition result;
Causing the computer to execute a display mode change procedure for changing the display mode of the portion related to the position information acquired in the search procedure;
In the display mode changing procedure, the display mode is changed so that a portion related to a character whose degree of coincidence with a character constituting the clear text string is lower than a predetermined value in the portion related to the position information can be identified. Clean book support program. A division procedure for dividing the clear character string into a plurality of partial character strings;
The cleanup support program according to claim 1, wherein the search procedure executes the search process for each partial character string and acquires the position information of the character string searched for each partial character string. A determination procedure for determining whether the character string searched for each partial character string is a continuous character string on the image based on the position information of the character string searched for each partial character string; ,
The cleanliness support program according to claim 2, wherein the display mode changing procedure changes the display mode when it is determined that the character string searched for each partial character string is a continuous character string on the image.   The cleanup support program according to any one of claims 1 to 3, wherein the search procedure executes the search processing using a part from the end of the clear text string.   5. The input candidate display procedure according to claim 1, further comprising: an input candidate display procedure for displaying, as a next input candidate, a character string obtained by character recognition for a predetermined range from the end of the portion related to the position information acquired in the search procedure. The written book support program described. A clean-up support method executed by a computer,
An image display procedure for causing the display device to display an image of the document data that is the target of the copy,
A clear character string input procedure for accepting input of a clear character string for the document data;
A search procedure for executing a search process using the clear character string as a keyword for the character recognition result for the image, and obtaining position information on the image of the searched character string from the character recognition result;
A display mode change procedure for changing the display mode of the portion related to the position information acquired in the search procedure,
In the display mode changing procedure, the display mode is changed so that a portion related to a character whose degree of coincidence with a character constituting the clear text string is lower than a predetermined value in the portion related to the position information can be identified. Clean book support method.

Images