JP2009048618A - Document extracting method, document extracting apparatus, computer program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document extracting method and a document extracting apparatus for extracting document data concerning a document composed of a plurality of pages from a database, and to provide a computer program and a recording medium. <P>SOLUTION: Document data corresponding to each page included in the document is stored, and furthermore, feature data indicative of a feature of the document data and a document index indicating the document are associated with the document data. A document extracting apparatus obtains input document data (S32), calculates feature data from the input document data (S34), judges similarity between the input document data and the document data based on the feature data (S36), obtains a document index associated with document data similar to the input document data (S39), and extracts a plurality of pieces of document data associated with the document index (S43). Thus, document data concerning the document including a page corresponding to the document data similar to the input document data is extracted for a plurality of pages. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a technique for retrieving a specific document from a database of documents, and more specifically, a document for retrieving document data corresponding to a read document from the database based on document data such as an image read by the scanner. The present invention relates to an extraction method, a document extraction device, a computer program, and a recording medium.

  Conventionally, data obtained by reading an original composed of a document or a photograph using a scanner, or original data created electronically using a personal computer (PC) or the like is stored in a database, and a new original is read and read. A technique for extracting original data corresponding to an original from a database is used. As a method for extracting manuscript data, for example, a keyword is extracted from a read manuscript using an OCR (Optical Character Reader), and the similarity of the manuscript is determined based on the keyword. For example, a method has been proposed in which the feature of a ruled line is extracted to determine the similarity of documents.

  Patent Document 1 associates a descriptor characterizing a document (document) with a list of documents characterized by the descriptor, generates a descriptor from the read document (input document), and collates the document using the generated descriptor. Techniques for performing are disclosed. The document descriptor is determined so as to be invariant to distortion or the like caused by the document reading. A plurality of descriptors are generated for one document, and the document associated with each of the descriptors generated from the scanned document is voted. The document with the highest number of votes or the number of votes exceeds a predetermined threshold. Select the original.

  In Patent Document 2, image data of a document is stored in advance, and pattern matching is performed in units of 1 bit between the scanned bitmap data and the stored bitmap data of the document. A technique for searching a manuscript is disclosed. Further, in Patent Document 2, in the case of a document composed of a plurality of pages, only the cover page is read for search, and the image data of the read page is compared with the stored first image data of each document. It is described that the manuscript may be searched.

In Patent Document 3, a document image is stored in advance, the feature amount of the read original image is compared with the feature amount of all pages of the stored document image, and the similarity is obtained. A technique for retrieving a document image by extracting a document image higher than a threshold value is disclosed. In this technique, when a plurality of document images are candidates, the document image is displayed and a selection by the user is accepted, and when the average similarity of pages included in the document image falls below a threshold value, the document is displayed. Delete images from candidates and refine them.
Japanese Patent Laid-Open No. 7-282088 JP-A-5-37748 JP 2006-31181 A

  Usually, a document such as a document is often composed of a plurality of pages. Although the conventional technique including the technique disclosed in Patent Document 1 can collate with a document read by a scanner and extract desired document data from a database, it is composed of a plurality of pages. It is necessary to extract original data by comparing each original. Therefore, when the original document to be collated is lost due to loss, dirt, or the like, there is a problem in that the original data relating to the original composed of a plurality of pages cannot be extracted over all pages. Patent Document 1 does not disclose any means for solving this problem.

  Further, in the technique for comparing bitmap data of documents composed of a plurality of pages as described in Patent Document 2, since comparison is performed for each page, the number of pages included in the document and the number of documents increase. There is a problem in that the comparison process takes time. In addition, when comparing bitmap data, it is necessary to accurately align two image data to be compared. However, in practice, it is difficult to accurately align the position, and as a result, there is a problem that a document cannot be searched with high accuracy.

  In the technique described in Patent Document 3, since the character code is extracted using the OCR as the feature amount in the character area of the document image, there is a problem that the accuracy of similarity determination is lowered depending on the extracted character code. is there. In order to compensate for this decrease in accuracy, it is conceivable to extract a large number of character codes, but in that case, the memory capacity for storing the character codes becomes large, and a search is performed using a lot of data. There is a problem that processing takes time. Further, in the techniques of Patent Documents 2 and 3, there is a problem that there is a possibility that a document including secret information may be easily output because no consideration is given to searching for a document including secret information.

  The present invention has been made in view of such circumstances, and an object of the present invention is to make it possible to extract data of other parts of a document based on a part of the document, thereby allowing a plurality of pages to be extracted. It is an object to provide a document extraction method, a document extraction device, a computer program, and a recording medium that can easily extract document data relating to a document composed of

  Another object of the present invention is to provide a document extraction device that can avoid the foolishness of erroneously extracting document data different from the purpose when document data is extracted.

  Another object of the present invention is to provide a document extraction device that can protect confidential information by setting conditions for outputting a document.

  The document extraction method according to the present invention is a method of extracting specific document data from document data stored in a storage means, and includes a document index indicating a document composed of a plurality of pages. Stored in association with the document data corresponding to each page by the storage means, calculated based on the feature points extracted from the document data, and storing the feature data indicating the characteristics of the document data in association with the document data To acquire input document data as new document data, extract feature points from the acquired input document data, and generate feature data indicating features of the input document data based on the extracted feature points Then, by comparing the generated feature data with the feature data stored in the storage means, it is related to the feature data stored in the storage means. The similarity between the document data and the input document data is determined, a document index associated with the document data determined to be document data having a high similarity with the input document data is acquired, and the document index indicated by the acquired document index is obtained. A plurality of document data corresponding to a plurality of contained pages is extracted.

  An original extracting apparatus according to the present invention includes original storing means for storing original data. In the original extracting apparatus for extracting specific original data from original data stored in the original storing means, a plurality of pages are used. Means for storing a document index indicating a document to be constructed in association with document data corresponding to each page included in the document, and a feature point extracted from the document data, and indicating characteristics of the document data Feature data storage means for storing feature data in association with the original data, acquisition means for acquiring input original data as new original data, and means for extracting feature points from the input original data acquired by the acquisition means And generating means for generating feature data indicating the characteristics of the input document data based on the feature points extracted by the means, and the generating means The similarity between the document data associated with the feature data stored in the feature data storage means and the input document data is determined by comparing the collected data with the feature data stored in the feature data storage means A document index associated with the document data determined by the determination unit, and a document indicated by the document index acquired by the unit. And an extraction means for extracting a plurality of document data corresponding to a plurality of contained pages.

  The document extraction device according to the present invention is configured such that the feature data storage means stores a plurality of feature data indicating characteristics of the document data in association with one document data, and the generation means includes: A plurality of feature data indicating features of input document data are generated, and the determination unit associates each of the plurality of feature data generated by the generation unit with feature data matching the feature data. A means for voting the received original data, and the original data having the maximum number of votes or the original data having a predetermined number or more of the original data stored in the original storage means, Means for determining that the document data has high similarity to the data.

  In the document extraction apparatus according to the present invention, the acquisition unit includes a unit that acquires a plurality of input document data, and the determination unit stores the document storage unit for each of the plurality of input document data. Means for determining the similarity between the document data and the input document data, and the extraction unit is configured to match document indexes associated with document data having a high similarity with each of the plurality of input document data. And a means for extracting a plurality of document data corresponding to a plurality of pages included in the document indicated by the document index.

  The document extracting apparatus according to the present invention is a document having a high similarity to each of a plurality of input document data when a plurality of document indexes associated with document data having a high similarity to the input document data are acquired. The apparatus further comprises means for requesting further input document data when a plurality of document indexes common to the plurality of input document data are obtained from the document indexes associated with the data.

  The document extraction device according to the present invention is characterized in that the acquisition means is configured to acquire input document data by optically reading the document.

  The document extracting apparatus according to the present invention stores, in association with the document index, a predetermined output condition necessary for outputting document data corresponding to each page included in the document indicated by the document index, and the extraction A means for determining whether or not an output condition associated with the document index associated with the document data extracted by the means is satisfied, and a document index when the output condition is determined to be satisfied Means for outputting a plurality of document data corresponding to a plurality of pages included in the document, and a plurality corresponding to a plurality of pages included in the document indicated by the document index when it is determined that the output condition is not satisfied And a means for prohibiting the output of the original data.

  The document extraction device according to the present invention is characterized by further comprising means for forming a plurality of images based on the plurality of document data extracted by the extraction means.

  A computer program according to the present invention is a computer program for causing a computer to extract specific document data from document data stored inside or outside the computer, and causes the computer to extract feature points from input document data. A procedure for causing the computer to generate feature data indicating the characteristics of the input document data based on the extracted feature points; and a computer for generating generated feature data and feature data indicating the characteristics of the stored document data. A procedure for determining the similarity between the stored document data and the input document data by comparing, and a document associated with the document data that the computer determines to be document data having a high similarity with the input document data The procedure to get the index and the computer Characterized in that it comprises a procedure for extracting a plurality of original data corresponding to a plurality of pages included in the document indicated by the index.

  The computer-readable recording medium according to the present invention can be read by a computer in which a computer program for extracting specific document data from document data stored inside or outside the computer is recorded. In a possible recording medium, a procedure for causing the computer to extract feature points from the input document data that has been input, and a procedure for causing the computer to generate feature data indicating the characteristics of the input document data based on the extracted feature points; A procedure for causing the computer to determine the degree of similarity between the stored document data and the input document data by comparing the generated feature data with the feature data indicating the characteristics of the stored document data; Related to manuscript data determined to be manuscript data with high similarity to data A computer program including a procedure for acquiring a document index and a procedure for extracting a plurality of document data corresponding to a plurality of pages included in a document indicated by the acquired document index is recorded on the computer. Features.

  In the present invention, document data corresponding to each page included in the document is stored, and further, feature data indicating the characteristics of the document data calculated based on the feature points extracted from the document data, and the document The document index shown is stored in association with the document data. When the original document data is acquired, the document extraction device generates feature data from the input document data, determines the similarity with the document data based on the feature data, and the document data having a high similarity with the input document data A document index associated with the document index is acquired, and a plurality of document data associated with the acquired document index is extracted. Thus, a document including pages corresponding to the document data determined to be similar to the input document data is specified, and document data corresponding to all pages included in the specified document is extracted.

  In the present invention, the document extraction device stores a plurality of feature data for one document data and determines the same for each feature data generated for the input document data in order to determine the similarity of the document data. Document data that has voted on document data associated with feature data and obtained a maximum number of votes or a number of votes equal to or greater than a predetermined amount is document data having a high degree of similarity to input document data. Since it is determined that the document data in which a lot of feature data matches among the plurality of feature data has a high similarity, it is possible to perform a more reliable similarity determination.

  In the present invention, the document extracting apparatus acquires a plurality of input document data, and associates the document index associated with the document data having a high degree of similarity with each input document data with the matched document index. The plurality of document data thus obtained is extracted. Thereby, it is possible to extract one original based on a plurality of pages.

  In the present invention, when there are a plurality of document indexes associated with document data having a high degree of similarity to the input document data, the document extraction device further requests input document data corresponding to other pages of the document. As a result, input document data corresponding to other pages of the document is further acquired, and the document index is narrowed down using the other pages of the document.

  In the present invention, the document extraction apparatus includes a scanner that optically reads a document as an acquisition unit that acquires input document data, and extracts document data by reading a part of the document with the scanner.

  Also, in the present invention, the document extraction device sets an output condition for each document index in advance, and outputs document data when the output condition is satisfied, and prohibits output of document data when the output condition is not satisfied. Thus, only the document corresponding to the document index that satisfies the output condition is output.

  Furthermore, in the present invention, the document extraction device includes means for forming an image based on the document data, so that an image based on the extracted document data can be formed.

  In the present invention, it is possible to extract document data corresponding to all pages of a document based on input document data corresponding to a part of a document composed of a plurality of pages. Therefore, even if a document composed of a plurality of pages is lost or lost due to dirt, document data covering all pages can be easily stored from a database in which document data is stored in advance. It becomes possible to extract.

  Further, according to the present invention, when determining the similarity of document data, it is possible to perform a more reliable similarity determination based on a plurality of feature data. It is possible to suppress erroneously determining that the document data is high.

  In the present invention, it is possible to extract one original based on a plurality of pages, and it is possible to further reduce the possibility of erroneously extracting original data different from the purpose. For example, even when there are similar documents, it is possible to extract target document data.

  Further, according to the present invention, by using a plurality of pages, it is possible to perform a more reliable similarity determination, and it is possible to extract desired document data with high accuracy.

  In the present invention, by scanning a part of a document with a scanner, for example, document data stored in a server device connected via a communication network can be extracted, and a photograph or document can be extracted. Thus, it is possible to easily acquire data of the entire original from a part of the original including the above.

  Further, according to the present invention, since the output of a document is possible when the output condition is satisfied, a document with high importance can be easily output by setting the output condition for a document with high importance. And the confidential information contained in the document can be protected.

  Further, in the present invention, an image forming apparatus such as a digital copying machine or a multifunction machine equipped with a scanner is used, and is connected to the image forming apparatus via document data stored in the image forming apparatus or a communication network. Since an image based on the document data extracted from the document data stored in the server device can be formed, it is possible to easily acquire a document composed of a photograph or a document by image formation. The invention has an excellent effect.

Hereinafter, the present invention will be specifically described with reference to the drawings showing embodiments thereof.
(Embodiment 1)
In the first embodiment, the document extracting apparatus of the present invention is an image forming apparatus that forms a color image. FIG. 1 is a block diagram showing an internal functional configuration of an original extracting apparatus 100 according to the first embodiment of the present invention. The document extraction apparatus 100 of the present invention includes a control unit 11 that controls the operation of each unit constituting the document extraction apparatus 100, a storage unit (storage unit) 12 including a semiconductor memory or a hard disk, and a color image optically. A color image input unit 13 for reading is provided. The color image input unit 13 is connected to a color image processing unit 2 that performs processing for generating image data corresponding to the read color image. The color image input unit 13 reads a manuscript consisting of a photograph or a document as a color image, and the storage unit 12 reads manuscript data which is image data generated by the color image processing unit 2 after the color image input unit 13 reads the manuscript. Remember. The storage unit 12 functions as a document storage unit in the present invention, and the color image input unit 13 functions as an acquisition unit in the present invention. The color image processing unit 2 is connected to a color image forming unit 14 that forms a color image based on the image data generated by the color image processing unit 2. An operation panel 15 that receives an operation from a user is connected to the color image input unit 13, the color image processing unit 2, and the color image forming unit 14.

  The color image input unit 13 is configured by a scanner having a CCD (Charge Coupled Device), and a reflected light image from a document which is a color image formed on a recording carrier such as paper is R (red) G. (Green) B (Blue) is decomposed and read by a CCD, converted into RGB analog signals, and output to the color image processing unit 2. The color image processing unit 2 performs image processing, which will be described later, on the RGB analog signals input from the color image input unit 13 to generate digital image data, and further, digital C (cyan) M (magenta) Y Image data composed of (yellow) K (black) signals is generated and output to the color image forming unit 14. Based on the image data input from the color image processing unit 2, the color image forming unit 14 forms a color image by a method such as thermal transfer, electrophotography, or inkjet. The operation panel 15 includes a display unit such as a liquid crystal display that displays information necessary for operation of the document extraction device 100, and a reception unit such as a touch panel or a numeric keypad that receives an instruction for controlling the operation of the document extraction device 100 by a user operation. It is comprised including.

  The color image processing unit 2 converts the analog signal input from the color image input unit 13 into a digital signal by the A / D conversion unit 20, and performs a shading correction unit 21, an input tone correction unit 22, a region separation processing unit 23, The document extraction processing unit 24, the color correction unit 25, the black generation and under color removal unit 26, the spatial filter processing unit 27, the output gradation correction unit 28, and the gradation reproduction processing unit 29 are sent in this order, and image data composed of digital CMYK signals. Is output to the color image forming unit 14.

  The A / D conversion unit 20 receives an RGB analog signal input from the color image input unit 13 to the color image processing unit 2, converts the RGB analog signal into a digital RGB signal, and converts the RGB signal into a shading correction unit 21. Output to.

  The shading correction unit 21 performs processing for removing various distortions generated in the illumination system, the imaging system, and the imaging system of the color image input unit 13 on the RGB signals input from the A / D conversion unit 20. Next, the shading correction unit 21 outputs the RGB signal from which distortion has been removed to the input tone correction unit 22.

  The input tone correction unit 22 adjusts the color balance for the RGB signals input from the shading correction unit 21. Further, the RGB signal input from the shading correction unit 21 to the input gradation correction unit 22 is an RGB reflectance signal, and the input gradation correction unit 22 converts the RGB signal input from the shading correction unit 21 into a color image. The signal is converted into a signal such as a density signal that can be easily processed by the processing unit 2. Next, the input tone correction unit 22 outputs the processed RGB signal to the region separation processing unit 23.

  The region separation processing unit 23 separates each pixel in the image represented by the RGB signal input from the input tone correction unit 22 into one of a character region, a halftone dot region, or a photo region, and based on the separation result, A region identification signal indicating to which region each pixel belongs is output to the black generation and under color removal unit 26, the spatial filter processing unit 27, and the gradation reproduction processing unit 29. The area separation processing unit 23 also outputs the RGB signal input from the input tone correction unit 22 to the document extraction processing unit 24.

  The document extraction processing unit 24 is connected to the storage unit 12 and relates to a process for inputting / outputting document data, which is image data composed of RGB signals, to / from the storage unit 12 and a document extraction method of the present invention described later. Execute the process. The document extraction processing unit 24 also outputs image data composed of RGB signals input from the region separation processing unit 23 or image data that is document data input from the storage unit 12 to the color correction unit 25. Note that the document extraction apparatus 100 may have a form in which the document extraction processing unit 24 is provided in parallel with the input tone correction unit 22 instead of being provided at the subsequent stage of the region separation processing unit 23.

  The color correction unit 25 converts the RGB signal input from the document extraction processing unit 24 into a CMY signal, and color turbidity based on the spectral characteristics of the CMY color material including unnecessary absorption components in order to realize faithful color reproduction. Is removed from the CMY signal. Next, the color correction unit 25 outputs the CMY signal subjected to color correction to the black generation and under color removal unit 26.

The black generation and under color removal unit 26 performs black generation processing for generating a K signal from the CMY three-color signals input from the color correction unit 25, and subtracts the K signal obtained by the black generation processing from the original CMY signal. Thus, the CMY three-color signal is converted into a CMYK four-color signal. As an example of the black generation process, there is a method of generating black using skeleton black. In this method, if the input / output characteristics of the skeleton curve are y = f (x), the data before conversion is C, M, Y, and the UCR (Under Color Removal) rate is α (0 <α <1), Data C ′, M ′, Y ′, and K ′ are expressed by the following equations.
K ′ = f (min (C, M, Y))
C ′ = C−αK ′
M ′ = M−αK ′
Y ′ = Y−αK ′

  Here, the UCR rate α (0 <α <1) indicates how much CMY is reduced by replacing the portion where CMY overlaps with K. The first equation indicates that the K signal is generated according to the smallest signal strength among the CMY signal strengths. Next, the black generation and under color removal unit 26 outputs the CMYK signal obtained by converting the CMY signal to the spatial filter processing unit 27.

  The spatial filter processing unit 27 performs spatial filter processing with a digital filter on the image represented by the CMYK signal input from the black generation and under color removal unit 26 based on the region identification signal input from the region separation processing unit 23. As a result, the blurring or graininess deterioration of the image is improved. For example, for a region separated into characters by the region separation processing unit 23, the spatial filter processing unit 27 uses a filter with a high enhancement amount of high-frequency components in order to improve character reproducibility. I do. Further, the spatial filter processing unit 27 performs low-pass filter processing for removing the input halftone dot component on the region separated into halftone dots by the region separation processing unit 23. Next, the spatial filter processing unit 27 outputs the processed CMYK signal to the output tone correction unit 28.

  The output gradation correction unit 28 performs an output gradation correction process for converting the CMYK signal input from the spatial filter processing unit 27 into a halftone dot area ratio that is a characteristic value of the color image forming unit 14, and The CMYK signal after the tone correction processing is output to the gradation reproduction processing unit 29.

  The gradation reproduction processing unit 29 reduces the number of pixel gradations based on the region identification signal input from the region separation processing unit 23 with respect to the CMYK signal input from the output gradation correction unit 28. Processing is performed so that the gradation corresponding to can be expressed. For example, for a region separated into characters by the region separation processing unit 23, the gradation reproduction processing unit 29 performs binarization or gradation reduction using a high-resolution screen suitable for reproducing high-frequency components. Perform the process. In addition, for a region separated into halftone dots by the region separation processing unit 23, the gradation reproduction processing unit 29 performs processing so that the image is finally separated into pixels and each gradation can be reproduced. Perform tone reproduction processing. Next, the gradation reproduction processing unit 29 outputs the processed image data to the color image forming unit 14.

  The color image forming unit 14 forms a CMYK color image on a record carrier such as paper based on the image data composed of CMYK signals input from the color image processing unit 2. By forming an image based on image data that is document data, the color image forming unit 14 outputs a document composed of a photograph or a document.

  Next, the configuration of the document extraction processing unit 24 and the processing performed by the document extraction processing unit 24 will be described. FIG. 2 is a block diagram showing the configuration of the document extraction processing unit 24. The document extraction processing unit 24 extracts feature points corresponding to characters or figures on the document represented by the input document data, and calculates feature data indicating the characteristics of the document data from the feature points. A data calculation unit 242; a voting processing unit 243 for voting on the document data stored in the storage unit 12 based on the feature data; a similarity determination processing unit 244 for determining the similarity of the document data based on the voting result; A document extraction unit 245 that extracts specific document data from the storage unit 12 is also provided.

  FIG. 3 is a block diagram showing a configuration of the feature point extraction unit 241. As shown in FIG. The feature point extraction unit 241 includes an achromatization processing unit 2410 for neutralizing the document data, a resolution conversion unit 2411 for converting the resolution of the document data to a predetermined resolution, and a filter processing unit 2412 for correcting the spatial frequency characteristics of the document data. , A binarization processing unit 2413 for binarizing document data, and a centroid extraction unit 2414 for extracting the centroid of characters and the like.

When the input document data is color image data, the achromatic processing unit 2410 achromatically converts the color image into a luminance signal or a brightness signal, and converts the converted document data to the resolution conversion unit 2411. Output. For example, the luminance signal Y is expressed as Yj = 0.30 × Rj + 0.59 × Gj + 0.11 × Bj, where Rj, Gj, and Bj are the intensity of the color component of each pixel RGB, and Yj is the luminance of each pixel. Can do. As another method, a method of achromatizing a color image by converting an RGB signal into a CIE (Commission International de l'Eclairage) 1976 L ^* a ^* b ^* signal may be used.

  The resolution conversion unit 2411 scales the document data so that the resolution of the input document data becomes a predetermined resolution, converts the resolution of the document data, and outputs the document data to the filter processing unit 2412. Thereby, even when the original is optically scaled by the color image input unit 13 and the resolution of the original data is changed, the feature points can be extracted without being affected by the change. The resolution conversion unit 2411 converts the resolution to a resolution lower than the resolution read at the same magnification by the color image input unit 13. For example, document data read at 600 dpi (dot per inch) by the color image input unit 13 is converted to 300 dpi. Thereby, the processing amount in the latter stage can be reduced.

  The filter processing unit 2412 corrects the spatial frequency characteristics of the input document data by image enhancement processing and smoothing processing, and outputs the corrected image to the binarization processing unit 2413. The processing in the filter processing unit 2412 is performed to absorb that the spatial frequency characteristics of the color image input unit 13 are different for each model. Image signals output by the CCD provided in the color image input unit 13 include optical system parts such as lenses or mirrors, aperture aperture of the light receiving surface of the CCD, transfer efficiency, afterimage, integration effect by physical scanning, scanning unevenness, etc. Due to this, the image becomes blurred. The filter processing unit 2412 restores the degradation that has occurred in the document data by performing an enhancement process such as a boundary or an edge. Further, the filter processing unit 2412 performs a smoothing process for suppressing high-frequency components that are not necessary for the feature point extraction process to be processed later.

  FIG. 4 is an explanatory diagram illustrating an example of a spatial filter used by the filter processing unit 2412. As shown in the figure, the spatial filter has a size of 7 × 7, for example, and is a mixed filter for performing enhancement processing and smoothing processing. The pixels of the input document data are scanned, and arithmetic processing using a spatial filter is performed on all the pixels. Note that the size of the spatial filter is not limited to 7 × 7, and may be 3 × 3, 5 × 5, or the like. In addition, the numerical values of the filter coefficients are examples, and are not limited thereto, and can be appropriately set according to the model or characteristics of the color image input unit 13.

  The binarization processing unit 2413 binarizes the document data by comparing the luminance value or brightness value of each pixel included in the input document data with a predetermined threshold value, and the binarized document data is converted into a centroid extraction unit. Output to 2414.

  The center-of-gravity extraction unit 2414 performs labeling for each pixel of the document data input from the binarization processing unit 2413, with a label corresponding to the binarized pixel value. That is, there are two types of labels, and when the pixel value is represented by 0 or 1, one label is attached to the 0 pixel and the other label is attached to the 1 pixel. Next, the center-of-gravity extraction unit 2414 specifies a connected region in which pixels with the same label are connected, extracts the center of gravity of the specified connected region as a feature point, and outputs the extracted feature point to the feature data calculation unit 242. To do. The feature points can be represented by coordinate values on the binary image represented by the document data.

  FIG. 5 is an explanatory diagram illustrating an example of feature points of a connected region. In FIG. 5, the identified connected area is the letter “A”, and is identified as a set of pixels with the same label. The position of the center of gravity of the letter “A” is a position indicated by a black circle in FIG. 5, and this center of gravity is a feature point. FIG. 6 is an explanatory diagram illustrating an example of a feature point extraction result for a character string. In the case of a character string composed of a plurality of characters, feature points are extracted at different positions depending on the character type. The feature points can be extracted not only for characters but also for a figure or a photograph part in the same manner. The feature point extraction method shown here is merely an example, and feature points may be extracted using other methods. For example, a process of decomposing a character string into words and extracting the centroid of each word as a feature point may be performed.

  The feature data calculation unit 242 performs a process of calculating feature data indicating the features of the input document data based on the feature points input from the feature point extraction unit 241. Here, a calculation example of feature data is shown. The feature data calculation unit 242 uses the feature points input from the feature point extraction unit 241 as target feature points in order, and extracts four other feature points close to the target feature point.

  FIG. 7 is an explanatory diagram showing the feature point of interest and the extracted feature points. As shown in FIG. 7, the feature data calculation unit 242 sets one feature point as a feature point of interest, and sets a predetermined number of feature points that are close to the periphery of this feature point of interest in order from the closest distance from the feature point of interest. Only (four points here) are extracted as peripheral feature points. In the example shown in FIG. 7, when the feature point a is the target feature point P1, the four feature points b, c, d, and e surrounded by the closed curve C1 in the figure are extracted as the peripheral feature points. When b is the target feature point P2, the four feature points a, c, e, and f surrounded by the closed curve C2 in the figure are extracted as the peripheral feature points.

  In addition, the feature data calculation unit 242 extracts a combination of three points from the four extracted peripheral feature points. FIG. 8 is an explanatory diagram illustrating an example in which feature data is calculated by extracting three peripheral feature points for the target feature point P1. As shown in FIGS. 8A to 8D, when the feature point a shown in FIG. 7 is the target feature point P1, three points are selected from the peripheral feature points b, c, d, and e. That is, all the combinations of the peripheral feature points b, c, d, the peripheral feature points b, c, e, the peripheral feature points b, d, e, and the peripheral feature points c, d, e are extracted.

  Next, the feature data calculation unit 242 calculates, for each extracted combination, an invariant (one of feature amounts) Hij with respect to geometric deformation. Here, i is a number indicating the feature point of interest (i is an integer equal to or greater than 1), and j is a number indicating a combination of three peripheral feature points (j is an integer equal to or greater than 1). In the present embodiment, the ratio of two of the lengths of the line segments connecting the peripheral feature points is set as the invariant Hij. Note that the length of the line segment may be calculated based on the coordinate value of each peripheral feature point. For example, in the example shown in FIG. 8A, the length of the line segment connecting the feature point b and the feature point c is A11, the length of the line segment connecting the feature point b and the feature point d is B11, The invariant H11 is obtained by H11 = A11 / B11. In the example shown in FIG. 8B, the length of the line segment connecting the feature point b and the feature point c is A12, the length of the line segment connecting the feature point b and the feature point e is B12, The invariant H12 is obtained by H12 = A12 / B12. In the example shown in FIG. 8C, the length of the line segment connecting the feature point b and the feature point d is A13, the length of the line segment connecting the feature point b and the feature point e is B13, The invariant H13 is obtained by H13 = A13 / B13. In the example shown in FIG. 8D, the length of the line segment connecting the feature point c and the feature point d is A14, the length of the line segment connecting the feature point c and the feature point e is B14, The invariant H14 is obtained by H14 = A14 / B14. In this way, invariants H11, H12, H13, and H14 are calculated in the examples shown in FIGS. 8 (a) to 8 (d). In the above example, the combination of the three neighboring feature points closest to the target feature point is j = 1, and the peripheral feature point 3 closest to the first, second, and fourth feature points is the target feature point. The combination of points is set to j = 2, the combination of 3 neighboring feature points closest to the first, third, and fourth feature points is set to j = 3, and the second, third, and fourth closest to the feature points of interest. The combination of 3 peripheral feature points was set to j = 4. Also, among the three peripheral feature points, Aij is a line segment connecting the peripheral feature point closest to the target feature point and the second closest peripheral feature point, and the third closest feature point to the peripheral feature point closest to the target feature point A line segment connecting the peripheral feature points is defined as Bij. In order to determine the order of the combination of the three peripheral feature points or the line segment used for calculating the invariant Hij, the length of the line segment connecting the peripheral feature points is not limited to the method used in the above example. What is necessary is just to define using arbitrary methods, such as the method defined based on.

Next, the feature data calculation unit 242 calculates a remainder value of the following formula as a hash value (feature data) Hi and stores it in the storage unit 12. Note that D in the following equation is a constant set in advance according to how much the range of values that the remainder can take is set.
(Hi1 × 10 ³ + Hi2 × 10 ² + Hi3 × 10 ¹ + Hi4 × 10 ⁰ ) / D

  Also, after the feature data calculation unit 242 finishes the extraction of the peripheral feature points for one target feature point and the calculation of the hash value Hi, the other feature point is set as the next target feature point, and the next target feature point is A feature point is extracted and a hash value is calculated, and a hash value is calculated with each feature point as a feature point of interest.

  In the example illustrated in FIG. 7, the feature data calculation unit 242 sets the feature point b as the target feature point P2 after the extraction of the peripheral feature points with the feature point a as the target feature point P1 and the calculation of the hash value H1 are completed. The extracted peripheral feature points and the hash value H2 are calculated. As shown in FIG. 7, when the feature point b is the target feature point P2, four feature points a, c, e, and f are extracted as the peripheral feature points. FIG. 9 is an explanatory diagram illustrating an example in which feature data is calculated by extracting three peripheral feature points for the target feature point P2. As shown in FIGS. 9A to 9D, the feature data calculation unit 242 includes a combination of three of the peripheral feature points a, c, e, and f, that is, the peripheral feature points a, e, and f. f, surrounding feature points a, c, e, surrounding feature points a, f, c, and surrounding feature points e, f, c are extracted, and an invariant Hij is calculated for each combination. As in the case of the feature point of interest P1 shown in FIG. 8, in the case of the feature point of interest P2, as shown in FIG. 9A, the invariant H21 is calculated by H21 = A21 / B21, and FIG. As shown in FIG. 9, the invariant H22 is calculated by H22 = A22 / B22, as shown in FIG. 9C, the invariant H23 is calculated by H23 = A23 / B23, and as shown in FIG. 9D, H24 = The invariant H24 is calculated by A24 / B24. The feature data calculation unit 242 calculates a hash value H2 from the invariants H21, H22, H23, and H24, and stores the hash value H2 in the storage unit 12. Further, the feature data calculation unit 242 repeats the same process using each feature point as the feature point of interest, and obtains and stores the hash value Hi when each feature point is the feature point of interest in the storage unit 12.

  As described above, the feature data calculation unit 242 calculates feature data that is the hash value Hi for each of the feature points, and sets the calculated feature data as feature data of the document data. The feature data calculation unit 242 functions as a generation unit in the present invention.

  Note that the feature data calculation method shown here is merely an example, and feature data may be calculated using another method. For example, the feature data may be calculated using another predetermined hash function. Further, when extracting feature points close to the feature point of interest, feature data may be calculated by extracting a number of feature points other than four, such as five or six points. In addition, three more feature points are extracted from the five extracted feature points, feature data is calculated based on the distance between the three points, and the number of features that can be extracted from the five feature points is the number of combinations. A process of calculating a plurality of feature data for one target feature point, such as calculating data, may be performed.

  The feature data calculated by the feature data calculation unit 242 is stored in the storage unit 12 in association with the document data. The storage unit 12 stores document data corresponding to each page for each document composed of a plurality of pages, and further associates a document table that associates document data with a document, and document data and feature data. A feature table is stored. The storage unit 12 functions as feature data storage means in the present invention.

  FIG. 10 is a conceptual diagram showing document data stored in the storage unit 12. A plurality of document data corresponding to each page included in the document is stored, and each document data has a page index of ID1, ID2,... Individually indicating the document data. FIG. 11 is a conceptual diagram illustrating an example of the contents of a document table that associates document data stored in the storage unit 12 with a document. A document index of Doc1, Doc2,... Individually indicating a document is recorded, and a page index indicating document data corresponding to each page included in the document is recorded in association with the document index. The table further records the number of pages of each document, and the same page index as the number of pages is associated with the document index. Since the page index is associated with the document index, the storage unit 12 stores the document index and document data in association with each other as shown in FIG.

  FIG. 12 is a conceptual diagram illustrating an example of the contents of a feature table that associates document data and feature data stored in the storage unit 12. In the figure, an example is shown in which feature data that is a hash value is calculated as E = 127. Each feature data of 0 to 126 is recorded, and the page index of the document data is recorded in association with the feature data calculated for the document data. Since the same feature data may be calculated for a plurality of document data, each feature data is associated with a plurality of page indexes. Further, since a plurality of feature data is calculated for one document data, the page index of one document data is associated with the plurality of feature data. Since the page index is associated with the feature data, the storage unit 12 stores the feature data and the document data in association with each other.

  The voting processing unit 243 searches the feature table stored in the storage unit 12 based on the feature data calculated by the feature data calculation unit 242, and the document indicated by the page index associated with the feature data that matches the calculated feature data Vote for data. When a plurality of page indexes are associated with one feature data, voting is performed for all document data associated with the feature data. Since the feature data calculation unit 242 calculates a plurality of feature data for the input document data, each feature data is voted, and a plurality of votes are performed for document data similar to the input document data. Is called. The voting processing unit 243 outputs the result of voting on the plurality of feature data calculated by the feature data calculating unit 242 to the similarity determination processing unit 244.

  The similarity determination processing unit 244 determines whether the input document data is similar to the document data stored in the storage unit 12 based on the voting result input from the voting processing unit 243, and the determination result Is output to the document extraction unit 245. Specifically, the similarity determination processing unit 244 inspects the number of votes of each document data stored in the storage unit 12, and the document data having the maximum number of votes is regarded as document data similar to the input document data. It is determined that Alternatively, the similarity determination processing unit 244 normalizes the number of votes obtained by dividing the number of votes of each document data by the maximum possible number of votes, which is the number of feature data calculated by the feature data calculation unit 242, and normalized number of votes Processing may be performed in which document data having a value equal to or greater than a predetermined threshold value is document data similar to the input document data. When there is document data similar to the input document data, the determination result output by the similarity determination processing unit 244 includes the page index of similar document data. The voting processing unit 243 and the similarity determination processing unit 244 function as determination means in the present invention.

  The document extraction unit 245 searches the document table stored in the storage unit 12 based on the page index included in the determination result input from the similarity determination processing unit 244, and acquires the document index associated with the page index. . Thus, a document including a page corresponding to the document data determined to be similar to the input document data is specified. Next, the document extraction unit 245 extracts a plurality of document data indicated by a plurality of page indexes associated with the acquired document index, and outputs the extracted plurality of document data to the color correction unit 25. Thereby, document data corresponding to all pages included in the specified document is extracted. The document extraction unit 245 functions as an extraction unit in the present invention.

  Next, processing executed by the document extraction apparatus 100 of the present invention having the above configuration will be described. The document extraction apparatus 100 executes a process of reading a document composed of a plurality of pages and registering document data, and a process of reading a part of the document and extracting document data corresponding to all pages of the document. . The process of extracting document data corresponding to all pages of the document from a part of the document is a process according to the document extraction method of the present invention. FIG. 13 is a flowchart showing a procedure of processing for registering document data.

  The control unit 11 of the document extracting apparatus 100 waits for reception of an instruction to register document data when the user operates the operation panel 15 (S11). When the registration instruction is not received (S11: NO), the control unit 11 continues to wait for the registration instruction. When a document data registration instruction is received (S11: YES), the user sets a document consisting of a plurality of pages on the document extraction device 100, and the color image input unit 13 optically reads each page. Then, a plurality of document data, which is image data composed of RGB signals, is acquired (S12). The color image input unit 13 outputs document data to the color image processing unit 2, and the color image processing unit 2 includes an A / D conversion unit 20, a shading correction unit 21, an input tone correction unit 22, and a region separation processing unit. The document data is processed in the order of 23, and the control unit 11 stores the document data in the storage unit 12 (S13).

  In the document extraction processing unit 24, the feature point extraction unit 241 extracts a plurality of feature points for one document data by the above-described processing (S14), and the feature data calculation unit 242 performs the feature for each feature point by the above-described processing. By calculating the data, a plurality of feature data indicating the features of one document data is calculated (S15). Next, the control unit 11 generates a page index indicating one document data, and sets the page index by adding the page index to the document data stored in the storage unit 12 (S16). At this time, the control unit 11 generates a unique page index based on the order in which the document data is input or the date and time. Next, the control unit 11 updates the feature table as shown in FIG. 12 by associating the feature data calculated by the feature data calculation unit 242 with the page index of the document data (S17).

  Next, the control unit 11 determines whether or not the process of associating the feature data with respect to all input document data has been completed (S18). If there is document data that has not yet been subjected to the process of associating the feature data (S18: NO), the control unit 11 returns the process to step S14, and the feature point extraction unit 241 has not yet extracted the feature point. Feature points are extracted from the document data. If the processing has been completed for all document data (S18: YES), the document index is set by generating a document index indicating a document composed of a plurality of pages corresponding to the plurality of acquired document data. (S19). Here, the control unit 11 generates a document index from the date and time. Note that the control unit 11 may perform a process of accepting the document index desired by the user through the operation panel 15.

  Next, the control unit 11 updates the document table stored in the storage unit 12 by associating the generated document index with the page index of the document data (S20), and ends the process. Through the above processing, document data of a document composed of a plurality of pages is stored in the storage unit 12.

  FIG. 14 is a flowchart showing a procedure of processing for extracting document data. The control unit 11 of the document extraction apparatus 100 waits for reception of an instruction to extract document data when the user operates the operation panel 15 (S31). When the extraction instruction is not received (S31: NO), the control unit 11 continues to wait for the extraction instruction reception. When an image data extraction instruction is received (S31: YES), the user sets some pages included in a document consisting of a plurality of pages to the document extraction apparatus 100, and the color image input unit 13 is set. By reading the page optically, input document data, which is image data composed of RGB signals, is acquired (S32).

  The color image input unit 13 outputs input document data to the color image processing unit 2, and the color image processing unit 2 includes an A / D conversion unit 20, a shading correction unit 21, an input tone correction unit 22, and a region separation process. The input document data is processed in the order of the unit 23, and in the document extraction processing unit 24, the feature point extraction unit 241 extracts a plurality of feature points from the input document data (S33). The feature data calculation unit 242 calculates feature data for each feature point extracted by the feature point extraction unit 241, thereby calculating a plurality of feature data indicating the features of the input document data (S34).

  Next, the voting processing unit 243 searches the feature table stored in the storage unit 12 for each feature data calculated by the feature data calculation unit 242, and votes for the document data indicated by the page index associated with the calculated feature data. A voting process is performed (S35). The similarity determination processing unit 244 determines which of the original data stored in the storage unit 12 is similar to the input original data based on the voting result in the voting processing unit 243 (S36). At this time, the similarity determination processing unit 244 selects the document data having the maximum number of votes from the document data having the minimum number of votes, or the document data having a normalized number of votes equal to or more than a predetermined threshold. Then, it is determined that the document data has high similarity to the input document data.

  Next, the control unit 11 determines whether or not the determination result in the similarity determination processing unit 244 indicates that there is document data having a high similarity (S37). When the determination result indicates that there is no document data having a high similarity (S37: NO), the control unit 11 does not have a document similar to the document read by the color image input unit 13 by the user. Is output (S38). Specifically, the control unit 11 displays character information indicating that there is no similar manuscript on the display unit of the operation panel 15, or displays an image that represents that there is no similar manuscript using characters. 14 to form. After step S38 is completed, the document extraction device 100 ends the process of extracting document data.

  If it is determined in step S37 that the determination result includes document data having a high degree of similarity (S37: YES), the document extraction unit 245 searches the document table stored in the storage unit 12 to determine the similarity. The processing unit 244 acquires a document index associated with the page index of the document data that is determined to have a high degree of similarity with the input document data (S39). Next, the control unit 11 determines whether or not a plurality of input document data corresponding to a plurality of pages has been acquired (S40). When the acquired input document data is input document data corresponding to one page (S40: NO), the document extraction unit 245 displays a plurality of pages indicated by a plurality of page indexes associated with the acquired document index in the document table. Document data is extracted (S43). As a result, all document data relating to a document including pages corresponding to document data having high similarity to the input document data is extracted.

  The document extraction unit 245 outputs the extracted document data to the color correction unit 25, and the color correction unit 25, the black generation and under color removal unit 26, the spatial filter processing unit 27, the output gradation correction unit 28, and the gradation reproduction processing unit. The document data is processed in the order of 29, and the color image processing unit 2 outputs the document data to the color image forming unit. The color image forming unit 14 performs an original output process of outputting an original composed of a plurality of pages corresponding to the plurality of original data by forming an image based on the plurality of original data as image data (S44). ). After step S44 is completed, the document extraction device 100 ends the process of extracting document data.

  When a plurality of input document data corresponding to a plurality of pages is acquired in step S40 (S40: YES), the control unit 11 determines whether or not the acquired document index matches each input document data. (S41). If the document indexes do not match (S41: NO), the control unit 11 advances the process to step S38 and outputs that there is no similar document.

  If the document indexes match in step S41 (S41: YES), the control unit 11 determines whether or not the processing for determining the similarity is completed for all input document data (S42). If there is input document data that has not yet been subjected to the process of determining the similarity (S42: NO), the control unit 11 returns the process to step S33, and the feature point extraction unit 241 still extracts the feature points. Feature points are extracted from input document data that has not been input. If processing has been completed for all input document data (S42: YES), document extraction apparatus 100 proceeds to step S43, and includes a page corresponding to document data having a high similarity to input document data. The document data relating to the document to be extracted is extracted and the document is output.

  In the above processing, it is assumed that there is one document data having a high similarity to the input document data. However, the document extraction apparatus 100 has document data whose normalized number of votes is equal to or greater than a predetermined threshold. When there are a plurality of document data, a process of determining that the plurality of document data is document data having a high similarity to the input document data may be performed. In this case, it is possible to perform processing for outputting a document related to each of a plurality of document data, and to display an image of a page corresponding to each document data determined to have a high degree of similarity as a display unit of the operation panel 15. May be displayed to allow the user to select valid document data.

  As described above in detail, in the present invention, the document extraction device 100 stores document data corresponding to each page included in the document in the storage unit 12, and further includes feature data indicating the characteristics of the document data, and the document. Is stored in association with the document data. When the document extraction apparatus 100 acquires input document data, the document extraction device 100 generates feature data from the input document data, determines a similarity with the document data based on the feature data, and a document with a high similarity with the input document data. A document index associated with the data is acquired, and a plurality of document data associated with the acquired document index is extracted. Thus, a document including pages corresponding to the document data determined to be similar to the input document data is specified, and document data corresponding to all pages included in the specified document is extracted. That is, it is possible to extract document data corresponding to all pages of a document based on input document data corresponding to a part of a document composed of a plurality of pages. Therefore, even if a document composed of a plurality of pages is lost due to loss or dirt, document data covering all pages of the document is stored in a database in which document data is stored in advance. It can be easily extracted.

  The document extraction apparatus 100 of the present invention stores a plurality of feature data for one document data and determines the same feature for each feature data generated for input document data in order to determine the similarity of document data. Document data that has voted for document data associated with the data and obtained the maximum number of votes or a number of votes equal to or greater than a predetermined amount is document data having a high degree of similarity to input document data. Since it is determined that the document data in which a lot of feature data matches among the plurality of feature data has a high similarity, it is possible to perform a more reliable similarity determination. Therefore, it is possible to avoid as much as possible the frustration of extracting document data different from the purpose by erroneously determining that document data not similar to the input document data is document data having a high degree of similarity.

  The document extraction device of the present invention acquires a plurality of input document data, and when document indexes associated with document data having a high degree of similarity with each input document data match, the document index is associated with the matched document index. A plurality of document data is extracted. Thereby, it is possible to extract a document based on a plurality of pages, and it is possible to further reduce the possibility of erroneously extracting document data different from the purpose. For example, even when there are similar documents, it is possible to reliably extract target document data.

  In the present invention, a feature point corresponding to the center of gravity of a character, a figure, a photograph, or the like on the document represented by the document data is extracted from the document data, and a numerical value is calculated based on the relative positional relationship of the extracted feature points. The feature data represented by is calculated. Since the document data is searched by comparing the feature data calculated in this way between the document data, the search is performed by comparing the conventional bitmap data, or a feature amount that is a large amount of character code extracted from the document. Compared with the search by comparing the document data, the amount of data necessary for performing the document data search process is greatly reduced. Therefore, in the present invention, the time required for the process of searching for document data is reduced as compared with the prior art. In the present invention, since document data is searched by comparing feature data obtained based on the relative positional relationship between a plurality of feature points, it is not necessary to perform image alignment between document data. Therefore, in the present invention, document data can be searched with higher accuracy than in the prior art.

  In the present embodiment, the document data that is color image data is handled. However, the present invention is not limited to this, and the document extraction apparatus 100 according to the present invention handles monochrome document data. Also good.

  In the present embodiment, the color image input unit 13 that is a scanner is used as the acquisition unit in the present invention. However, the present invention is not limited to this, and the document extraction device 100 of the present invention is used as the acquisition unit. Alternatively, an interface for receiving document data from an external scanner or PC may be provided. The document data according to the present invention is not limited to image data obtained by optically capturing a document, and may be application data such as text data created by a PC using an application program. In this case, the document extraction apparatus 100 receives document data that is application data through an interface that is an acquisition unit, and executes processing according to the present invention.

  In the present embodiment, the acquired document data is registered, and necessary document data is extracted from the registered document data. However, the present invention is not limited to this. The extraction apparatus 100 may be configured to perform processing for extracting document data without performing registration processing by a method such as attaching a storage unit 12 that stores document data in advance. Further, in the present embodiment, an embodiment has been described in which processing for extracting necessary document data from document data stored in the storage unit 12 built in the document extraction apparatus 100 is performed, but the present invention is not limited to this. The document extraction device 100 may be configured to perform processing for extracting necessary document data from document data stored in an external storage unit such as a storage device or a server device connected via a communication network.

(Embodiment 2)
In the second embodiment, when there are a plurality of document data having a high similarity to the input image data, the input image data is further acquired to narrow down the image data. The internal configuration of the document extracting apparatus according to the present embodiment is the same as that of the first embodiment described with reference to FIGS. In addition, the storage contents in the storage unit 12 according to the present embodiment are the same as those in the first embodiment described with reference to FIGS. The process of registering document data by the document extraction apparatus according to the present embodiment is the same as that in the first embodiment described with reference to the flowchart of FIG.

  15 and 16 are flowcharts showing a procedure of processing for extracting document data performed by the document extraction apparatus according to the second embodiment. The control unit 11 of the document extraction apparatus 100 waits for reception of an instruction to extract document data when the user operates the operation panel 15 (S501). If no extraction instruction has been received (S501: NO), the control unit 11 continues to wait for an extraction instruction to be received. When an image data extraction instruction is received (S501: YES), the user sets some pages included in a document consisting of a plurality of pages to the document extraction apparatus 100, and the color image input unit 13 is set. By reading the one page optically, input document data, which is image data composed of RGB signals, is acquired (S502).

  The color image input unit 13 outputs input document data to the color image processing unit 2, and the color image processing unit 2 includes an A / D conversion unit 20, a shading correction unit 21, an input tone correction unit 22, and a region separation process. The input document data is processed in the order of the unit 23, and in the document extraction processing unit 24, the feature point extraction unit 241 extracts a plurality of feature points from the input document data (S503). The feature data calculation unit 242 calculates feature data for each feature point extracted by the feature point extraction unit 241 to calculate a plurality of feature data indicating the features of the input document data (S504).

  Next, the voting processing unit 243 searches the feature table stored in the storage unit 12 for each feature data calculated by the feature data calculation unit 242, and votes for the document data indicated by the page index associated with the calculated feature data. A voting process is performed (S505). The similarity determination processing unit 244 determines which of the original data stored in the storage unit 12 is similar to the input original data based on the voting result in the voting processing unit 243 (S506). In step S506, the similarity determination processing unit 244 determines that the document data whose normalized vote count is equal to or greater than a predetermined threshold is document data having a high similarity to the input document data.

  Next, the control unit 11 determines whether or not the determination result in the similarity determination processing unit 244 indicates that there is document data having a high similarity to the input document data (S507). If the determination result indicates that there is no document data having a high similarity (S507: NO), the control unit 11 does not have a document similar to the document read by the color image input unit 13 by the user. Is output (S508). After step S508 is completed, the document extraction apparatus 100 ends the process of extracting document data.

  If it is determined in step S507 that the determination result includes document data having a high similarity to the input document data (S507: YES), the document extraction unit 245 searches the document table stored in the storage unit 12. Then, the document index associated with the page index of the document data determined by the similarity determination processing unit 244 to have a high similarity with the input document data is acquired (S509). If there are a plurality of document data having a high degree of similarity with the input document data, a plurality of document indexes are acquired in step S509. Next, the control unit 11 determines whether or not the input document data currently being processed is input document data obtained by reading the second and subsequent pages of a plurality of pages (S510). When the input document data currently being processed is input document data obtained by reading the first page of the document (S510: NO), the control unit 11 determines whether or not there are a plurality of document indexes acquired in step S509. (S515). If there is a single document index acquired in step S509 (S515: NO), the document extraction unit 245 extracts a plurality of document data indicated by a plurality of page indexes associated with the acquired document index in the document table ( S516).

  The document extraction unit 245 outputs the extracted document data to the color correction unit 25, and the color correction unit 25, the black generation and under color removal unit 26, the spatial filter processing unit 27, the output gradation correction unit 28, and the gradation reproduction processing unit. The document data is processed in the order of 29, and the color image processing unit 2 outputs the document data to the color image forming unit. The color image forming unit 14 performs an original output process of outputting an original composed of a plurality of pages corresponding to the plurality of original data by forming an image based on the plurality of original data as image data (S517). ). After step S517 is completed, the document extraction apparatus 100 ends the process of extracting document data.

  In step S510, when the input document data currently being processed is input document data obtained by reading the second and subsequent pages of the document (S510: YES), the control unit 11 corresponds to the pages read so far from the document. It is determined whether or not there is a document index common to all the pages read so far among the document indexes acquired for the input document data to be input (S511). If there is no document index common to all pages (S511: NO), the control unit 11 advances the process to step S508, and outputs that there is no similar document.

  If there is a document index common to all the pages read so far (S511: YES), the control unit 11 determines whether there are a plurality of document indexes common to all pages (S512). When the document index common to all pages is singular (S512: NO), the control unit 11 proceeds to the processing unit step S516, and the document extraction unit 245 has a plurality of page indexes associated with the acquired document index. The color image forming unit 14 performs document output processing for outputting a document composed of a plurality of pages corresponding to the plurality of document data (S517), and the document extraction device 100 extracts the document data shown in FIG. Ends the process.

  When there are a plurality of document indexes acquired in step S515 (S515: YES), or when there are a plurality of document indexes common to all the pages read so far (S512: YES), the control unit 11 Performs a process of outputting information indicating a request for another page of the document (S513). Specifically, the control unit 11 causes the display unit of the operation panel 15 to display character information that requests reading of a new page included in the document.

  Next, the control unit 11 determines whether or not another page included in the document is set in the document extraction apparatus 100 by the user (S514). When another page included in the document is set in the document extraction apparatus 100 (S514: YES), the control unit 11 returns the process to step S502, and the color image input unit 13 determines other pages included in the document. Input document data corresponding to the page is acquired.

  If another page included in the document is not set in the document extraction apparatus 100 (S514: NO), the control unit 11 advances the process to step S516. Note that in step S514, the control unit 11 causes the document to be set when another page of the document is not set even after a predetermined time has elapsed after the process of step S513 is completed, or when the user operates the operation panel 15. When a reading end instruction is received, processing for determining that no other page of the document is set may be performed. When the control unit 11 advances the process to step S516, the document extraction unit 245 extracts document data indicated by each page index associated with each of a plurality of document indexes common to all the pages read so far (S516). The color image forming unit 14 performs a document output process for outputting a document corresponding to the extracted document data (S517). As a result, the document extraction device 100 outputs a plurality of documents corresponding to a plurality of document indexes. After step S517 ends, the document extraction apparatus 100 ends the process.

  As described above in detail, the document extraction apparatus according to the present embodiment has a plurality of document indexes associated with document data having high similarity to input document data corresponding to pages read in the document. Input document data corresponding to another page of the document is requested, and input image data obtained by reading another page of the document is acquired. Furthermore, the document extraction apparatus according to the present embodiment acquires a document index associated with document data having a high similarity to the input document data in common for all read pages, and a plurality of document indexes associated with the acquired document index. Original data is extracted. As a result, when there are a plurality of document indexes of document data determined to be similar to the input document data, the document index is narrowed down using other pages of the document, and the document data similar to the input document data is retrieved. The narrowing is repeated until the document index is determined. Therefore, by using a plurality of pages, it is possible to perform a more reliable similarity determination, and it is possible to extract desired document data with high accuracy.

(Embodiment 3)
In the first and second embodiments, a form in which any original can be output based on the input original data corresponding to one page has been shown. However, in the third embodiment, the output conditions for a specific original are more stringent. Shows the form. The internal configuration of the document extracting apparatus according to the present embodiment is the same as that of the first embodiment described with reference to FIGS.

  FIG. 17 is a conceptual diagram illustrating a content example of a document table that associates document data and a document stored in the storage unit 12 according to the third embodiment. A page index and the number of pages are recorded in association with the document indexes of Doc1, Doc2,... That individually indicate the document, and further, output conditions necessary for outputting the document are recorded in association with the document index. Yes. In the example shown in FIG. 17, output conditions are not associated with the document indexes of Doc1 to Doc4, and output conditions are associated with the document indexes of Doc21 and Doc51. The document index of Doc 21 is associated with an output condition that both the document data corresponding to ID 21 and ID 25 among the page indexes ID 21 to ID 28 associated with the document index are similar to the input document data. The document index of Doc 51 is associated with an output condition that document data corresponding to three or more page indexes among ID 51 to ID 55 associated with the document index is similar to the input document data. Yes. The contents of the feature table that associates the document data and the feature data stored in the storage unit 11 according to the present embodiment are the same as those in the first embodiment described with reference to FIG.

  The process of registering document data by the document extraction apparatus according to the present embodiment is the same as that in the first embodiment described with reference to the flowchart of FIG. The document data extraction process performed by the document extraction apparatus according to the present embodiment is the same as that in the first embodiment described with reference to the flowchart of FIG. 14 or the embodiment described with reference to FIGS. Although it is almost the same as in the case of 2, the content of the document output process in step S44 or step S517 is different from that in the first or second embodiment.

  FIG. 18 is a flowchart illustrating a procedure of document output processing performed by the document extraction apparatus according to the third embodiment. The document extraction apparatus 100 according to the present embodiment executes steps S31 to S43 shown in FIG. 13 or steps S501 to S516 shown in FIGS. 14 and 15 in the process of extracting document data. In the document output process of step S44 or step S517, the control unit 11 first selects a document index associated with one document data among the document data extracted by the document extraction unit 245 in step S43 or step S516 ( S61). Next, the control unit 11 searches the document table stored in the storage unit 12 and determines whether an output condition is associated with the selected document index (S62). When the output condition is associated with the selected document index (S62: YES), the control unit 11 determines whether the output condition associated with the document index is satisfied (S63).

  For example, when the document index of Doc 21 shown in FIG. 17 is selected, when document data corresponding to ID21 and ID25 are both determined as document data similar to the input document data in step S37 or step S507. It is determined that the output condition is satisfied. If the document data corresponding to either ID 21 or ID 25 is not determined to be document data similar to the input document data, it is determined that the output condition is not satisfied. If the document index of Doc21 is selected, document data corresponding to three or more page indexes among the page indexes of ID51 to ID55 are document data similar to the input document data in step S37 or step S507. When it is determined, it is determined that the output condition is satisfied. When only document data corresponding to less than three page indexes is determined as document data similar to input document data, it is determined that the output condition is not satisfied.

  When the output condition is not associated with the document index in step S62 (S62: NO), or when the output condition associated with the document index is satisfied in step S63 (S63: YES), the color image forming unit 14 Forms an image based on the document data indicated by each page index associated with the selected document index, and outputs a document corresponding to the selected document index (S64). For example, the documents corresponding to the document indexes Doc1 to Doc4 shown in FIG. 17 are output unconditionally because the output conditions are not defined. Documents corresponding to the document indexes of Doc21 and Doc51 are output when the output conditions are met. After step S64 is completed, the control unit 11 advances the process to the next step S65. If the output condition associated with the document index is not satisfied in step S63 (S63: NO), the control unit 11 proceeds to the next step S65 without outputting the document corresponding to the selected document index. Proceed. In this way, the control unit 11 prohibits output of document data that does not satisfy the output condition.

  Next, the control unit 11 determines whether or not the processing for all document data extracted in step S43 or step S516 has been completed (S65). If there is document data that has not been processed yet (S65: NO), the control unit 11 returns the process to step S61, and stores the document index associated with the document data extracted in step S43 or step S516. Select a document index that has not yet been selected. When the processing for all the document data extracted in step S43 or step S516 is completed (S65: YES), the control unit 11 ends the document output processing and returns the processing to the processing for extracting the document data. After the document output process is completed, the document extraction apparatus 100 ends the process of extracting document data.

  As described above in detail, the document extraction apparatus according to the present embodiment sets output conditions for each document index in advance, and when performing document output processing, only the document corresponding to the document index that satisfies the output conditions. Is output. In Embodiments 1 and 2, since a document can be output based on input document data corresponding to one page, even a highly important document including confidential information is based on one page of the document. Thus, all manuscript pages can be easily output. In the present embodiment, the document extraction device outputs a document for which an output condition is determined when the output condition is satisfied, so by setting the output condition for a highly important document, It is possible to prevent a highly important document from being easily output.

  For example, as an output condition, it is necessary to determine that input document data and document data are similar in a plurality of pages, so that all pages of a highly important document based on one page of the document Can be prevented from being output. Further, since it is necessary to determine that the input document data and the specific document data are similar as output conditions, a user who does not own a specific page of the document can extract the document from the document extraction device. Can no longer be extracted from. As specific manuscript data, manuscript data representing contents for collation not related to the main contents of a manuscript composed of a plurality of pages may be registered. It is more desirable that the collation contents have a completely different format from the main contents of the manuscript, such as when the main content of the manuscript is Japanese, the collation content is English.

  As described above, the document extraction apparatus according to the present embodiment extracts a document having an output condition for a specific user who owns specific document data for collation. It is possible to prevent other users who do not have specific document data for collation from outputting highly important documents. Therefore, in the present embodiment, it is possible to protect confidential information included in a document by setting output conditions for a highly important document including confidential information.

(Embodiment 4)
In the first to third embodiments, the document extracting apparatus of the present invention is an image forming apparatus. In the fourth embodiment, the document extracting apparatus of the present invention is a scanner apparatus. FIG. 19 is a block diagram showing an internal functional configuration of the document extracting apparatus 300 according to the fourth embodiment of the present invention. The document extraction apparatus 300 of the present invention includes a control unit 31 that controls the operation of each unit constituting the document extraction apparatus 300, a storage unit 32 that includes a semiconductor memory or a hard disk, and a color image input that optically reads a color image. A portion 33 is provided. An A / D conversion unit 34 is connected to the color image input unit 33, a shading correction unit 35 is connected to the A / D conversion unit 34, and a document extraction processing unit 36 is connected to the shading correction unit 35. Yes. The document extraction processing unit 36 is connected to a transmission unit 37 that transmits document data to the outside. The storage unit 32, the color image input unit 33, the A / D conversion unit 34, the shading correction unit 35, the document extraction processing unit 36, and the transmission unit 37 are connected to the control unit 31. An operation unit 38 that receives an operation from the user is connected.

  The storage unit 32 stores document data corresponding to each page for each document composed of a plurality of pages, similarly to the storage unit 12 included in the document extraction apparatus 100 described in the first to third embodiments. Further, a document table that associates document data and a document, and a feature table that associates document data and feature data are stored. The transmission unit 37 is connected to an external PC or an image forming apparatus.

  The color image input unit 33 is configured by a scanner equipped with a CCD. The reflected light image from the original is decomposed into RGB, read by the CCD, converted into RGB analog signals, and sent to the A / D conversion unit 34. Output. The A / D conversion unit 34 converts RGB analog signals into digital RGB signals, and outputs the RGB signals to the shading correction unit 35.

  The shading correction unit 35 performs processing for removing various distortions generated in the illumination system, the imaging system, and the imaging system of the color image input unit 33 on the RGB signal input from the A / D conversion unit 34. Further, the shading correction unit 35 performs processing for adjusting the color balance of the RGB signals, and performs processing for converting the RGB reflectance signals into density signals. Next, the shading correction unit 35 outputs document data, which is image data composed of processed RGB signals, to the document extraction processing unit 36.

  The document extraction processing unit 36 is configured in the same manner as the document extraction processing unit 24 provided in the document extraction apparatus 100 described in the first to third embodiments, and executes the same processing as the document extraction processing unit 24. That is, the document extraction processing unit 36 uses the document data input from the shading correction unit 35 as input document data and performs the same processing as the processing shown in the flowcharts of FIGS. A plurality of document data relating to a document including pages corresponding to document data having a high similarity to the input document data is extracted from 32.

  The control unit 31 outputs the extracted document data by causing the transmission unit 37 to transmit a plurality of document data extracted by the document extraction processing unit 36 to the outside. The transmission unit 37 transmits a plurality of document data to an apparatus such as an external PC or an image forming apparatus, and the external apparatus executes processing such as forming an image based on the plurality of document data.

  As described above in detail, in the present embodiment as well, in the same manner as in the first to third embodiments, all pages of a document are supported based on input document data corresponding to a part of a document composed of a plurality of pages. Document data can be extracted. Therefore, also in the present embodiment, even when a document composed of a plurality of pages is lost due to loss or dirt, all the documents of the document are stored in the database in which the document data is stored in advance. Document data over a page can be easily extracted.

(Embodiment 5)
In the fifth embodiment, a form in which the document extraction apparatus of the present invention is realized using a general-purpose computer will be described. FIG. 20 is a block diagram showing the internal configuration of the document extraction device 400 according to the fifth embodiment of the present invention. A document extraction apparatus 400 according to the present embodiment is configured using a general-purpose computer such as a PC, and includes a CPU 41 that performs a calculation, a RAM 42 that stores temporary information generated along with the calculation, A drive unit 43 such as a CD-ROM drive for reading information from the recording medium 5 of the present invention such as an optical disk and a storage unit 44 such as a hard disk are provided. The CPU 41 causes the drive unit 43 to read the computer program 51 of the present invention from the recording medium 5 of the present invention, and stores the read computer program 51 in the storage unit 44. The computer program 51 is loaded from the storage unit 44 to the RAM 42 as necessary. Based on the loaded computer program 51, the CPU 41 executes processing necessary for the document extraction device 400.

  The document extracting apparatus 400 includes an input unit 45 such as a keyboard or a pointing device for inputting information such as various processing instructions operated by the user, and a display unit 46 such as a liquid crystal display for displaying various information. It has. Further, the document extraction device 400 includes a transmission unit 47 connected to an external output device 61 that outputs a document such as an image forming device, and a reception unit connected to an external input device 62 that inputs document data such as a scanner device. 48. The transmission unit 47 transmits the document data to the output device 61, and the output device 61 outputs the document based on the document data. The input device 62 optically reads a document to generate document data, transmits the generated document data to the document extraction device 400, and the receiving unit 48 receives the document data transmitted from the input device 62. The receiving unit 48 functions as an acquisition unit in the present invention.

  The storage unit 44 stores document data corresponding to each page for each document composed of a plurality of pages, similarly to the storage unit 12 included in the document extraction apparatus 100 described in the first to third embodiments. Further, a document table that associates document data and a document, and a feature table that associates document data and feature data are stored.

  The CPU 41 loads the computer program 51 of the present invention into the RAM 42, and executes processing according to the document extraction method of the present invention in accordance with the loaded computer program 51. That is, when document data is input from the input device 62 by the receiving unit 48, the CPU 41 uses the input document data as input document data, and performs the same processing as that shown in the flowchart of FIG. 14, FIG. 15, or FIG. Thus, a plurality of document data relating to a document including pages corresponding to document data having a high similarity to the input document data is extracted from the storage unit 44. The CPU 41 transmits a plurality of extracted document data from the transmission unit 47 to the output device 61, and the output device 61 outputs a document composed of a plurality of pages based on the document data. Note that the CPU 41 may perform processing for handling application data such as text data created using an application program as document data.

  As described above in detail, in the present embodiment as well, in the same manner as in the first to fourth embodiments, all pages of a document are supported based on input document data corresponding to a part of a document composed of a plurality of pages. Document data can be extracted. Therefore, also in the present embodiment, even when a document composed of a plurality of pages is lost due to loss or dirt, all the documents of the document are stored in the database in which the document data is stored in advance. Document data over a page can be easily extracted.

  In the present embodiment, an embodiment has been described in which processing for extracting necessary document data from document data stored in the storage unit 44 built in the document extraction apparatus 400 is performed, but the present invention is not limited to this. The document extraction device 400 may be configured to perform processing for extracting necessary document data from document data stored in an external storage unit (not shown) such as a storage device or a server device connected via a communication network.

  The recording medium 5 of the present invention in which the computer program 51 of the present invention is recorded is a magnetic tape, a magnetic disk, a portable hard disk, an optical disk such as a CD-ROM / MO / MD / DVD, or an IC card (memory). Any type of card-type recording medium such as an optical card) may be used. Further, the recording medium 5 of the present invention is mounted on the document extraction device 400, and a semiconductor memory from which the CPU 41 can read the recorded contents of the recording medium 5, that is, mask ROM, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically). An Erasable Programmable Read Only Memory), a flash ROM, or the like may be used.

  The computer program 51 of the present invention is downloaded to the document extraction device 400 from an external server device (not shown) connected to the document extraction device 400 via a communication network such as the Internet or a LAN and stored in the storage unit 44. Form may be sufficient. In the case of this form, a program necessary for downloading the computer program 51 is stored in the storage unit 44 in advance, or is read from a predetermined recording medium using the drive unit 43 and stored in the storage unit 44. It is sufficient if it is stored and loaded into the RAM 42 as necessary.

FIG. 3 is a block diagram showing an internal functional configuration of the document extraction device according to the first embodiment of the present invention. 3 is a block diagram illustrating a configuration of a document extraction processing unit. FIG. It is a block diagram which shows the structure of a feature point extraction part. It is explanatory drawing which shows the example of the spatial filter which a filter process part utilizes. It is explanatory drawing which shows the example of the feature point of a connection area | region. It is explanatory drawing which shows the example of the extraction result of the feature point with respect to a character string. It is explanatory drawing which shows an attention feature point and the extracted feature point. It is explanatory drawing which shows the example which extracts the 3 surrounding feature points with respect to the attention feature point P1, and calculates feature data. It is explanatory drawing which shows the example which extracts 3 surrounding feature points with respect to the feature point of interest P2, and calculates feature data. It is a conceptual diagram which shows the document data which a memory | storage part memorize | stores. It is a conceptual diagram which shows the example of the content of the manuscript table which matches the manuscript data and manuscript which a memory | storage part memorize | stores. It is a conceptual diagram which shows the example of the content of the feature table which matches the original data and feature data which a memory | storage part memorize | stores. 6 is a flowchart illustrating a procedure of processing for registering document data. It is a flowchart which shows the procedure of the process which extracts original data. 10 is a flowchart illustrating a procedure of processing for extracting document data performed by the document extraction device according to the second embodiment. 10 is a flowchart illustrating a procedure of processing for extracting document data performed by the document extraction device according to the second embodiment. 10 is a conceptual diagram illustrating an example of the contents of a document table that associates document data and a document stored in a storage unit according to Embodiment 3. FIG. 10 is a flowchart illustrating a procedure of document output processing performed by a document extraction device according to Embodiment 3. FIG. 10 is a block diagram showing an internal functional configuration of an original extracting apparatus according to a fourth embodiment of the present invention. FIG. 10 is a block diagram illustrating an internal configuration of a document extraction device according to a fifth embodiment of the present invention.

Explanation of symbols

100, 300, 400 Document extraction device 11, 31 Control unit 12, 32, 44 Storage unit (storage unit)
13, 33 Color image input unit 24, 36 Document extraction processing unit 242 Feature data calculation unit 243 Voting processing unit 244 Similarity determination processing unit 245 Document extraction unit 41 CPU
5 Recording medium 51 Computer program

Claims (10)

In a method for extracting specific document data from document data stored in a storage means,
A document index indicating a document composed of a plurality of pages is stored in a storage unit in association with document data corresponding to each page included in the document,
Feature data that is calculated based on the feature points extracted from the document data and indicates the features of the document data is stored in association with the document data in a storage unit,
Input document data that is new document data is acquired,
Extract feature points from the acquired input document data,
Based on the extracted feature points, generate feature data indicating the features of the input document data,
By comparing the generated feature data with the feature data stored in the storage means, the similarity between the document data associated with the feature data stored in the storage means and the input document data is determined,
Obtain a document index associated with document data determined to be document data having a high degree of similarity to the input document data,
A document extraction method comprising: extracting a plurality of document data corresponding to a plurality of pages included in a document indicated by an acquired document index. In a document extraction device comprising document storage means for storing document data, and extracting specific document data from document data stored in the document storage means,
Means for storing a document index indicating a document composed of a plurality of pages in association with document data corresponding to each page included in the document;
Feature data storage means for storing the feature data calculated based on the feature points extracted from the document data and indicating the features of the document data in association with the document data;
Acquisition means for acquiring input original data which is new original data;
Means for extracting feature points from the input document data acquired by the acquisition means;
Generating means for generating feature data indicating the characteristics of the input document data based on the feature points extracted by the means;
Document data associated with the feature data stored in the feature data storage means and input document data by comparing the feature data generated by the generation means with the feature data stored in the feature data storage means Determining means for determining the similarity to
Means for obtaining a document index associated with the document data determined by the determination means as document data having a high similarity to the input document data;
An original extracting apparatus comprising: extraction means for extracting a plurality of original data corresponding to a plurality of pages included in the original indicated by the original index acquired by the means. The feature data storage means includes
A plurality of feature data indicating characteristics of the document data are stored in association with one document data;
The generating means includes
It is configured to generate a plurality of feature data indicating the features of the input document data,
The determination means includes
A means for voting on the document data associated with the feature data matching the feature data for each of the plurality of feature data generated by the generation means;
Among the document data stored in the document storage unit, the document data having the maximum number of votes or the document data having a number of votes equal to or more than a predetermined amount is determined to be document data having high similarity to the input document data. The document extracting device according to claim 2, further comprising: The acquisition means includes
Means for acquiring a plurality of input document data;
The determination means includes
Means for determining the similarity between the original data stored in the original storage means and the input original data for each of a plurality of input original data;
The extraction means includes
When document indexes associated with document data having high similarity with each of the plurality of input document data match each other, a plurality of document data corresponding to a plurality of pages included in the document indicated by the document index is extracted. The document extracting device according to claim 2 or 3, further comprising: means. When a plurality of document indexes associated with document data having a high degree of similarity with input document data is acquired, or among document indexes associated with document data having a high similarity with each of the plurality of input document data 5. The document extracting apparatus according to claim 4, further comprising means for requesting further input document data when a plurality of document indexes common to the plurality of input document data are acquired. The acquisition means includes
6. The document extraction device according to claim 2, wherein input document data is acquired by optically reading a document. Means for storing predetermined output conditions necessary for outputting document data corresponding to each page included in the document indicated by the document index in association with the document index;
Means for determining whether an output condition associated with a document index associated with the document data extracted by the extraction unit is satisfied;
Means for outputting a plurality of document data corresponding to a plurality of pages included in a document indicated by a document index when it is determined that the output condition is satisfied;
The apparatus further comprises means for prohibiting output of a plurality of document data corresponding to a plurality of pages included in a document indicated by a document index when it is determined that the output condition is not satisfied. The document extraction device according to any one of 2 to 6. The document extracting apparatus according to claim 2, further comprising a unit that forms a plurality of images based on the plurality of document data extracted by the extracting unit. In a computer program for causing a computer to extract specific document data from document data stored inside or outside the computer,
A procedure for causing a computer to extract feature points from input document data;
A procedure for causing the computer to generate feature data indicating the features of the input document data based on the extracted feature points;
A procedure for causing the computer to determine the degree of similarity between the stored document data and the input document data by comparing the generated feature data with the feature data indicating the characteristics of the stored document data;
A procedure for causing a computer to acquire a document index associated with document data determined to be document data having high similarity to input document data;
A computer program comprising: causing a computer to extract a plurality of document data corresponding to a plurality of pages included in a document indicated by an acquired document index. In a computer-readable recording medium in which a computer program for extracting specific document data from document data stored inside or outside the computer is recorded in a computer,
A procedure for causing a computer to extract feature points from input document data;
A procedure for causing the computer to generate feature data indicating the features of the input document data based on the extracted feature points;
A procedure for causing the computer to determine the degree of similarity between the stored document data and the input document data by comparing the generated feature data with the feature data indicating the characteristics of the stored document data;
A procedure for causing a computer to acquire a document index associated with document data determined to be document data having high similarity to input document data;
A computer-readable recording comprising a computer recorded with a computer program including a procedure for extracting a plurality of document data corresponding to a plurality of pages included in a document indicated by an acquired document index Medium.

Images