JPH06290251A - Document image processing method of electronic file device - Google Patents

Abstract

PURPOSE:To display image data while masking only a secret item entry area even if the document format is different by making the secret item entry area of the image data according to coordinate values and displaying the data other than the area in the state of the original image. CONSTITUTION:A main memory MM stores the program of a microprocessor and also stores image data, etc., outputted from a scanner SCN temporarily. An optical disk OD stores with image data which are inputted from the scanner SCN and compressed by an image processing processor IPU and document indexes converted into text data by a character recognition part ORC, i.e., key words for document retrieval correlatingly. For registration, the document structure is analyzed to extract tfi entry item of a secret item, whose coordinate values are stored on the optical disk OD together with the image data. Then, when a document image is used, an image mask pattern corresponding to position information on the secret item entry area is generated to display or print a document image wherein, the area is hidden.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic file device for automatically registering a large number of documents by recognizing an index from a document image. In particular, the present invention relates to a document image processing method suitable for displaying an image in which a confidential item is masked in correcting a misrecognized index for a document having a confidential item.

[0002]

2. Description of the Related Art As a conventional technique for automatic document registration, for example, in the "Document Understanding Method" described in Japanese Patent Laid-Open No. 61-59568, frame information is extracted from a document image to create format information. A method of reading characters on a document using format information is disclosed. Also, JP-A-60-1016
In "Character reading method" described in Japanese Patent Publication No. 81, there is disclosed a technology for increasing recognition accuracy of a document having various character types by controlling the recognition method by using the format information described for each character entry field. ing.

Regarding security protection, for example, the "faring system" described in Japanese Patent Laid-Open No. 61-285578.
Discloses a method in which a specific pattern such as halftone dot is superimposed and output on the entire surface of the document image.

[0004]

However, in order to accurately search the automatically registered documents, it is necessary to correct the recognition result. This correction work increases the number of registered documents and diversifies the document format. Along with this, there are many cases where it is performed outside the document management department. Here, for documents having various formats, it is necessary to display the image of the index writing area clearly and the image of the confidential item writing area with masking (covering).

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an electronic file device for automatically registering a large number of documents by recognizing an index from a document image, and in particular, confidentiality. An object of the present invention is to provide a document image processing method for masking and outputting a confidential item description area for a document having an item. Further, another object of the present invention is to provide a document image processing method for masking and outputting only the confidential item area in correcting an index that is erroneously recognized.

[0006]

A document image processing method according to the present invention is a method for automatically registering a tabular document represented by a form in an electronic file device, and a method for displaying the registered document. .

Specifically, a scanner for photoelectrically converting a document image on paper, a first memory for temporarily storing the converted image, and a document format are identified from the image information stored in the memory. And a document structure analysis unit for extracting the index entry area and the confidential item entry area, and reading the image information of the index entry area from the first memory, recognizing a character pattern and converting it into a text code, and recognizing the pattern. An electronic filing device comprising a character recognition unit for outputting the similarity of the above, and a large-capacity second memory for storing a large number of document images and their indexes, and (a) is stored in the first memory. Document image, the position information of the index entry area and confidential item entry area extracted by the document structure analysis, and the character recognition index. (B) When using the stored document image, the image mask pattern corresponding to the position information of the confidential item description area is generated, When displaying or printing a document image with the area covered, or when correcting the index using the document image stored in (b) ', the document data whose similarity is equal to or less than the preset value is stored in the second memory. It is characterized in that the image mask pattern corresponding to the position information of the confidential item description area is generated by searching and reading, and the document image covering the area is displayed or printed.

[0008]

The typical operation of the document image processing method according to the present invention is as follows. First, in registration, the document structure is analyzed using the format information created in advance, the confidential item description area is extracted, and the coordinate values thereof are stored in the second memory together with the image data.

Next, when using the document image, an image mask pattern corresponding to the position information of the confidential item description area is generated, and the document image with the area covered is displayed or printed.

In the display for index correction, the coordinate values of the confidential item description area stored in the second memory are read, an image mask pattern corresponding to the area is generated, and the image data and the mask pattern are combined. After performing a logical operation, display it on the display. By this method, even if the format of the document to be registered is different, it is possible to display the image in which only the confidential area is covered.

[0011]

Embodiments of the present invention will now be described in detail with reference to the drawings. First, FIG. 8 shows a format example of a document to be automatically registered in an electronic file device (also referred to as an electronic filing device). The figure is an example of a remuneration statement, and the ruled lines Lx, L
In the areas F1, F2, F4, and Fn surrounded by y and the like, document indexes, in other words, items f1, f2, f4, and fn that are keywords for document search are described. Specifically, f1 is "Compensation statement 1993 3
Date such as "month", f2 is the personal identification number of the payee, f4
Is the number of working days, and fn is the total amount. In the area F3, a confidential item f3 that should not be shown to anyone other than the person concerned with the payment of rewards
(Name of payee) is listed. Further, the present remuneration statement is characterized in that the position and size of the area F are slightly different for each issuing business, and the typeface of the item f is different.

Next, the configuration of an electronic filing apparatus having a document automatic registration function to which the present invention is applied will be described with reference to FIG. In the figure, SCN indicates an image scanner which optically reads the document of the format shown in FIG. 8 with a pixel number of 200 or 400 dots per inch and binarizes 1 bit per pixel to output an image. . SCC is
A control unit for connecting the scanner SCN and the bus by an instruction of a microprocessor μP described later is shown.

Reference numeral MM denotes a main memory, which stores a microprocessor program and temporarily stores image data output from the scanner. The MM is composed of a 4 MB (megabyte) semiconductor memory.

FOR indicates a document structure analysis unit, which analyzes characteristics such as ruled line lengths and start point coordinates of image data input from a scanner, and collates with the format information stored in advance on the magnetic disk DK to input document. Format is identified and the coordinate value of the index description area is calculated. OCR
Indicates a character recognition unit, which cuts out image data based on the coordinate values of the index writing area obtained by the FOR,
The character pattern in the area is recognized and converted into text data.

IPU stands for Image Processor,
The binary image read from the main memory MM is converted into MMR.
Compress using encoding method and transfer to optical disc,
On the contrary, the compressed image is decompressed and sent to the display bitmap memory, the pattern for partially masking (covering) the image is generated, or the decompressed image data and the mask pattern are ORed. Has image processing function.

OD indicates an optical disk having a storage capacity of 600 MB per sheet and its drive unit, and the scanner S
The image data input from the CN and compressed by the image processor IPU is associated with the document index converted into text data by the character recognition unit OCR, that is, the keyword for document search is stored. The ODC is an optical disk control unit that controls reading / writing of the optical disk OD according to a command from the microprocessor μP, and can control a maximum of 100 optical disks. Since the document image data read by the scanner SCN is efficiently compressed, it is possible to register about 30,000 documents on one optical disc.

The microprocessor μP has a keyboard K
The scanner SCN, the document structure analysis unit FOR, the character recognition unit OCR, the image processing processor IPU, the optical disc OD, and the like are controlled in accordance with the user's processing request input from B. BUS is a 32-bit wide data bus.
CRT is a device operation guidance, document index,
And a display for displaying a document image. BM
Is a bitmap memory in which image data to be displayed on the CRT is stored. DC denotes a display control unit that synthesizes the display data of the microprocessor μP and the image data output from the image processing processor IPU and writes the combined data in the bitmap memory BM. KB is a keyboard for inputting processing requests to the apparatus, search conditions, etc., and KBC is a keyboard control unit.

Next, FIG. 10 shows the structure of the keyword information written in a predetermined sector of the optical disc when the document is registered. The figure shows the keyword information per document, and the information length K is 512B (bytes). A document index, that is, a keyword can be attached to some items (item K1, item K2, ..., Item Kn). In the case of a remuneration statement, examples of these items include date, personal number, number of days worked, total amount, and the like. In the keyword configuration information field P,
The number of items, each item name, and the data length of each item are written, and in each item field Q, a keyword output from the character recognition unit OCR is written. Optical disc management information field R
In, the address where the document image data is stored and the image data length are written.

A document image automatic registration, correction display, and search display method according to the present invention will be described with reference to FIGS. The feature is that in registration, the coordinate values of the confidential item description area obtained by the document structure analysis are stored on the optical disk together with the image data and the index recognized by the character recognition unit, and in the correction / search display, the image read from the optical disk is stored. By appropriately masking the confidential item description area, the confidential portion can be properly displayed.
Hereinafter, each registration, correction, and search method will be described in detail.

(1) Automatic Document Registration Method First, FIG. 6 shows format information. The format information is necessary to identify the format of the input document and obtain the coordinate value of the area in which each item is described. The figure shows the method to describe the format quantitatively. The area in which the document index is written and the area in which confidential items are written (Xij,
Yij), ΔXij, ΔYij, and the attribute of each item is aij, sij
Described in. Where (Xij, Yij): the value of the starting point coordinates of the ruled line forming the area Fi ΔXij: the length in the X direction of the ruled line forming the area Fi ΔYij: the length in the Y direction of the ruled line forming the area Fi aij: When the character type of the item fi described in the area Fi is machine-written numbers, aij = 0 machine-written kanji, aij = 1 handwritten numbers, aij = 2 handwritten kanji, aij = 3 (machine and Is a word processor or the like) sij: sij = 0 when the item fi described in the area Fi is not a confidential item, and sij = 1 when the item is a confidential item. The subscript j represents a format number. this is,
This is because the format, that is, the position and size of the area Fi, and the typeface of the characters of the item fi are different for each issuing establishment of the remuneration statement.

In this way, as shown in FIG. 7, format information is prepared in advance for several formats (Format 1 to m) and prepared as a format information file on the magnetic disk.

Next, a method of registering a document image will be described with reference to FIG. In the figure, when a "document registration" request is input from the keyboard KB, the microprocessor μP sets the input document number j to 1 in step JINT, and activates the scanner in the next step SCN. The scanner optically reads the document image in the procedure IMG and binarizes it. The μP writes the image data imgj to the address M1 on the main memory MM.

Next, in step FOR, the document structure analysis unit analyzes the features such as the length of the ruled line and the starting point coordinates of the image data imgj, and inputs them by collating them with the format information TFOR prepared in advance on the magnetic disk DK. The document format is identified, and the coordinate data (Xim, Yim), ΔXim, and ΔYim (these are collectively referred to as secij) of the item description area of the corresponding format m, and the item attribute data aim, sim.
(Referred to as aij and sij, respectively) are transferred to the M2 address of the main memory.

The specific method of format identification is as follows: a) Track consecutive black pixels existing in the input image imgj and extract a closed ruled line frame. b) The starting point coordinates (Xk, Yk) of the ruled line frame and the lengths ΔXk, Δ
Find Yk. c) From the format information file ((Xim, Yim), ΔXim, Δ
Yim) is read. d) For each format, ((Xk, Yk), ΔXk, ΔYk) is associated with ((Xim, Yim), ΔXim, ΔYim). That is, for all regions i

[0025]

## EQU1 ## min [(Xim-Xk) ² + (Yim-Yk) ² ] (Equation 1) k which satisfies k (k = 1 to n) is obtained, and ((Xk, Yk), ΔXk of the region k is obtained. , ΔYk)
Is replaced by ((Xi, Yi), ΔXi, ΔYi). e) Next, for each format

[0026]

[Equation 2]

[0027]

[Equation 3]

[0028]

## EQU4 ## ΔPm = ΔNm + ΔLm (Equation 4) is calculated. f) The format m that minimizes ΔPm in Equation 4 is the same as the format of the input document.

G) ((Xim, Yim), ΔXim, Δ of format m
Yim), aim, and sim are read from the format information file and transferred to the main memory.

Next, in step SEC, the character recognition unit s
If ij = 0 (area i is not a confidential item), the above FOR
Coordinate data sec of the index entry area obtained by
Image data in the range specified by ij is stored in M1 of the main memory.
Read out from the address (cut out the image) and recognize the character pattern in the image by the procedure OCR to detect the text data ind.
ij (character recognition). At the same time, the OCR also outputs the pattern matching similarity likij.

In the character recognition, the item attribute data a
It is possible to improve recognition accuracy by selectively driving the optimum recognition method based on ij.

In the procedure FIL, the image data imgj is read from the address M1 in the main memory, compressed by the image processor IPU, and then stored (filing) in the optical disc OD. Coordinate data seci of each item description area
The information sij indicating whether the item is a confidential item or not is also stored in the optical disc in the same manner as the image data. The index text data indij which is the output of the above-mentioned OCR and the similarity l
ikij is a keyword field of the optical disc (see FIG.
At 0, write to K1, K2, ..., Kn).

Note that steps IINT, IEND, and IINC indicate that image clipping of a predetermined number of areas and character recognition processing are repeated for one document. Also, the procedures JINT, JEND, and JINC.
Returns S for all documents set in the scanner.
It indicates that a series of registration processing from CN to FIL is repeated.

(2) Document Index Correction Display Method Generally, in automatic registration, a character pattern in the index writing area is recognized, converted into text data, and written as a search keyword on the optical disk. But the character 1
Since it is very difficult to recognize it with the accuracy of 00%, it is necessary to correct the recognition result in order to perform a highly accurate search. The correction is performed by displaying the original image and the recognition result at the same time, comparing the two, and inputting the correct character from the keyboard for misrecognized characters, but the correction work is performed because the number of documents to be registered is huge. In many cases, a specialized department other than the management department is requested, and it is necessary to mask confidential items and display them as images.

An image display method for correcting the document index (search keyword) will be described with reference to FIG. In the same figure, after inputting a "correction start" request from the keyboard KB, in step THIN, the corrector changes the KB from the similarity threshold Lth.
Enter. This is to select a document that needs to be corrected by utilizing the fact that the index having a higher possibility of recognition error has a lower similarity.

Next, in step SRH, the similarity lik is calculated from the keyword field (Ki in FIG. 10) of the optical disk.
read ij,

[0037]

## EQU00005 ## Documents such as likij.ltoreq.Lth ... (Equation 5) are searched.

In the procedure PAT, by referring to the optical disc management information field R of the keyword information, s
The coordinate data secij of the area where ij = 1 (area where confidential items are written) is read from the optical disc OD, and a mask pattern having a value of "0" only in the area portion is generated.

In steps MSK and DSP, the image data imgj of the entire document is read from the OD, expanded by the image processor, and the mask pattern is superimposed and displayed on the display.

Specifically, the original image after Imgj is expanded,
Mij is mask pattern (secij, that is, (Xim,
Yim), ΔXim, and ΔYim are defined by Mij
= 0, and Mij = 1) in the range other than

[0041]

[Equation 6] Dspj = Imgj∧Mij (Equation 6) is calculated (∧ represents a logical product operator), and Dspj is displayed. If there are multiple confidential areas,

[0042]

[Equation 7] Dspj = ((Imgj∧M1j) ∧M2j) ∧M3j ... (Equation 7) is displayed.

An example of a display screen according to the above procedure is shown in FIG. In the figure, CRT indicates a display screen, and WIN indicates a window for displaying an image. DOC is the result of displaying the document image Dspj by the operation of Equation 6. In the case where the confidential item is described in the area F3, the area F3 is masked and the character f3 is covered and displayed.

Although omitted here, in another window, the result (indij) of recognizing the items f1, f2, f4, and fn by the OCR and the corrected character string input from the KB are displayed.

In the procedure REV of FIG. 2, the corrector corrects the document index when there is a recognition error while looking at the screen of FIG. In the procedure ENT, the corrected index is written again in the keyword field of the OD (the old index is rewritten).

Procedures JINT, JEND, and JIN
By C, a series of correction processing from SRH to ENT is repeated until a document satisfying likij ≦ Lth does not exist in the OD.

(3) Document Retrieval Display Method An image display method in document retrieval will be described with reference to FIG. In the figure, after inputting a "search start" request from the keyboard KB, the user inputs the password PW from the KB in step PWIN.

In the procedure TMIN, the user inputs a search term (search condition) TM from the KB.

Next, in the procedure SRH, the keyword field of the optical disk (K1, K2, ..., Kn in FIG. 10).
Documents that have an index that matches the search term TM, that is,

[0050]

## EQU00008 ## Documents indij = TM ... (Equation 8) are searched.

In the procedure PCHK, the password PW entered in PWIN is compared with the password PWo registered in advance in the electronic filing device as a searcher. If they match, the document image is displayed by the procedure IMG and the procedure DSP. A display screen according to this procedure is shown in FIG. In the figure, the document image DOC displays the entire area without masking the confidential area.

If PW ≠ PWo, the procedure PAT, MS
By K and DSP, the confidential area is masked and displayed as in FIG. The method is the same as the correction display described above.

In step SEND, if the end of search is not input from KB, the process returns to step TWIN and the next document search process is performed.

In the above embodiment, the remuneration statement is taken as a document to be automatically registered in the electronic filing apparatus and corrected and retrieved, and the formats (position of the item description area, size, etc.) are slightly different. Although the example of the document is explained, since the format information is created corresponding to the document to be registered, if it is a tabular document, documents with different formats such as remuneration statement, resume, work table, etc. will be mixed. May be.

The coordinate value of the confidential item description area stored in the main memory and used for the image mask is not the value of the format information file, but the document structure analysis value (procedure b described in FIG. 1) ((Xk , Yk), ΔXk, ΔYk)). Further, it may be the sum of the area based on the value of the format information file and the area based on the document structure analysis value.

[0056]

As described above, according to the present invention, in automatically registering a tabular document in the electronic filing apparatus, displaying the correction, and searching, in registration, the confidential item description extracted by the document structure analysis is described. The coordinate values of the area are stored on the optical disk together with the image data and the index recognized by the character recognition unit, and in the correction / search, the confidential item description area of the image data read from the optical disk is masked based on the coordinate value, and the area Other than that, since it is displayed in the state of the original image, even if the document format is different, only the confidential item description area can be masked and displayed.

[Brief description of drawings]

FIG. 1 is a diagram showing a document image registration method according to an embodiment of the present invention.

FIG. 2 is a diagram showing an image display method for correcting a document index in the embodiment.

FIG. 3 is a diagram showing an image display method in a document search according to the embodiment.

FIG. 4 is a diagram showing a display screen of a document image in which a confidential area is masked in the embodiment.

FIG. 5 is a diagram showing a screen displaying the entire area of the document image in the embodiment.

FIG. 6 is a diagram showing format information in which a format in the embodiment is quantitatively described.

FIG. 7 is a diagram showing a format information file in the embodiment.

FIG. 8 is a diagram showing a format example of a document to be automatically registered by the electronic filing device.

FIG. 9 is a diagram showing a configuration of an electronic filing apparatus having a document automatic registration function to which the present invention is applied.

FIG. 10 is a diagram showing a structure of keyword information of an optical disc in an electronic filing device to which the present invention is applied.

[Explanation of symbols]

SCN ... Scanner, FOR ... Document structure analysis unit, OCR ...
Character recognition unit, IPU ... Image processor, OD ... Optical disc, DK ... Magnetic disc, MM ... Main memory, μ
P ... Microprocessor, Fi ... Index or confidential item entry area, fi ... Index or confidential item,
secij ... Index or confidential item description area coordinate data, aij ... Index description area attribute, sij ...
Confidential item identification information, imgj ... document image data, ind
ij ... Character recognition index, likij ... Character recognition similarity, DOC ... Document image, MSK ... Image mask, T
M ... Search term, Ki ... Keyword.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Eiichi Hatano 1-280 Higashi Koigokubo, Kokubunji, Tokyo Inside Central Research Laboratory, Hitachi, Ltd. (72) Inventor Masaaki Fujinawa 2880, Kozu, Odawara-shi, Kanagawa Hitachi, Ltd. Storage System Division

Claims (4)

[Claims] 1. A scanner for photoelectrically converting a document image on a paper surface, a first memory for temporarily storing the converted image, a document format based on the image information stored in the memory, and an index. A document structure analysis unit that extracts the description area and confidential item description area, a character recognition unit that reads the image information of the index description area from the first memory, recognizes a character pattern, and converts it into a text code, and a large number of documents. It has an image and a large-capacity second memory for storing its index, and registers a tabular document,
An electronic file device for outputting a registered document, comprising: (a) the document image stored in the first memory, the position information of the confidential item description area extracted by the document structure analysis,
The text code of the character recognition index is stored in association with the second memory, and (b) when the stored document image is used, the display of the document image in which the area corresponding to the position information of the confidential item description area is covered Alternatively, a document image processing method for an electronic file device, which comprises printing. 2. An operation mode for displaying or printing the entire area of a stored document image and an operation mode for displaying or printing a document image in which a confidential item description area is covered by a password input by a user of the electronic file device. 2. The document image processing method for an electronic file device according to claim 1, wherein 3. A scanner for photoelectrically converting a document image on paper, a first memory for temporarily storing the converted image, a document format identified from the image information stored in the memory, and an index. A document structure analysis unit that extracts the description area and the confidential item description area, reads the image information of the index description area from the first memory, recognizes the character pattern and converts it into a text code, and the degree of similarity of the pattern recognition. It is equipped with a character recognition unit that outputs a, a large-capacity second memory that stores a large number of document images and their indexes, and registers a tabular document,
An electronic file device for outputting a registered document, comprising: (a) the document image stored in the first memory, the position information of the index entry area and the confidential item entry area extracted by the document structure analysis, and character recognition. The text code of the index is stored in the second memory in association with the similarity, and (b) when correcting the index using the stored document image, the document image having the similarity equal to or less than a preset value is stored in the second memory. A document image processing method for an electronic file device, comprising: searching and reading from a memory, displaying or printing a document image in which an area corresponding to the position information of the confidential item description area is covered and hidden. 4. The electronic file device according to claim 3, wherein (a) when the similarity of recognition is equal to or more than a preset threshold, that is, when the reliability of recognition is high, the text code of the recognized index is Stored in the second memory,
If the recognition similarity is less than or equal to the threshold value, the document image stored in the first memory, the position information of the index entry area and confidential item entry area extracted by the document structure analysis, and the text code of the recognized index are displayed. Stored in association with the second memory, and (b) when correcting the index using the stored document image, display or print the document image in which the area corresponding to the position information of the confidential item description area is covered. A document image processing method for an electronic file device, comprising: