JP2021081853A - Document reader - Google Patents

Abstract

To provide a document reader which allows for easily finding out transfer of an image for certification such as an impression of a seal due to ink transfer.SOLUTION: In the document reader, a page inside specification unit 102 detects an image group of an impression of a seal where vermilion has a high frequency and characters used in the seal are included, from each of pieces of page data. An inter-page consistency determination unit 105 evaluates whether the image group of the impression of the seal is included in each of a pair of pieces of page data corresponding to two document surfaces adjacent to each other in a bundle of documents, whether the image group of the impression of the seal included in one of the pair and that included in the other have a mirror image relation, and which of them is degraded. If it is evaluated as a result of evaluation that the image group of the impression of the seal in either of the pair of pieces of page data is degraded, a user is notified of the result of evaluation.SELECTED DRAWING: Figure 2

Description

The present invention relates to a document scanning apparatus that reads a bundle of documents composed of a plurality of documents and obtains scan data composed of a plurality of page data, and more particularly to an improvement in image processing for the page data.

In recent years, document readers have been introduced in many workplaces in order to perform dual management of important documents using paper and electronic data. The document reading device feeds out a bundle of documents loaded on a tray one by one by an ADF (Automatic Document Feeder), reads the documents, and creates document data. Further, it is possible to set the double-sided mode / single-sided mode and the mixed setting of the document size. When the double-sided mode is set, the document data corresponding to the pages in which the front and back sides of the document are continuous can be obtained. The document data obtained by scanning with the document reading device may include an imprint formed by imprinting the document. Here, as a prior art that extracts an image portion of a seal stamp from print data and executes some processing, the image forming system described in Patent Document 1 is known. Such an image forming system stores an electronic seal stamp image in advance, and extracts the seal stamp image from the print data together with the electronic seal stamp image. Then, it is determined whether or not the extracted seal stamp image matches any of the stored electronic seal stamp images, and if the extracted seal stamp image matches any one or a plurality of electronic seal stamp images, printing is performed. To execute.

Japanese Unexamined Patent Publication No. 2010-44563

When the originals are loaded on the ADF tray and used for automatic scanning, there is a possibility that ink transfer may occur between the originals. Ink transfer means that the imprint of the imprint is transferred to the paper surface of another original because the ink, which is the stamp, is superimposed on another original without being dried. It is assumed that in a bundle of documents placed on a tray of a document reading device, a paper stamped with a real seal is used as an underlay for documents to be shipped outside the company. As a result, ink is transferred from the paper to the document, and when the imprint of the actual seal is transferred from the paper to the document, the important imprint of the actual seal is leaked to the outside. And if the transcript leaked outside the company is used for forgery of the seal, it may lead to serious legal troubles.

With the increase in the capacity of the built-in memory, a document reading device capable of reading 100 or more documents at a time is currently introduced in many business establishments. It is a heavy burden for the clerk to visually check the presence or absence of transfer of each manuscript from among 100 or more manuscripts that can be read at one time. Further, if such visual confirmation is thoroughly performed, there is a problem that the efficiency of office work is reduced at once.

Although the image forming system of Patent Document 1 can extract the stamp image from the print data, it cannot determine whether the extracted stamp image is due to the ink image, so that the ink image is as described above. There is a problem that the transfer of the imprint cannot be found.

Although the transfer of the imprint by imprint has been described as an example, the above-mentioned problem is not limited to the transfer of the imprint. A universal proof image that can occur when a proof image is transferred from one document to another, which is a proof that the transfer occurs when the tray of the document reader is placed and is responsible for the description of the document. It is a technical issue.

An object of the present disclosure is to provide a document reader capable of easily finding a transfer of a proof image due to ink transfer.

The above-mentioned problem is a document reading device that reads a plurality of loaded double-sided originals and obtains page data corresponding to each of both sides of each original, and has two original surfaces corresponding to two facing original surfaces in the state of the document bundle. It is evaluated whether or not one of the page data contains a proof image that proves that the manuscript is responsible for the description, and the other contains an image that has a mirror image relationship with the proof image. The problem is solved by a document reading device including an evaluation means and a notification means for notifying that a certification image has been transferred between documents based on the evaluation result.

Here, the evaluation means is a process of calculating a color histogram of each page data, and an extraction process of extracting a distribution area of color components having a frequency lower than the distribution area of the main color components of the manuscript body from the color histogram. , And the pixel group consisting of the color components of the distribution area extracted by the extraction process may be used as the proof image.

The proof image is an image of an imprint formed by imprinting a manuscript, and the evaluation means is based on the color component of the pigment used for the imprint and / or the color component of a color in the vicinity thereof. An image for proof may be obtained by extracting a group of pixels. The proof image is an image of an imprint formed by imprinting a stamp on a manuscript, and the evaluation means obtains a frame shape of the imprint. The process of extracting may be performed.

Even if there is a break in the edge forming the peripheral edge of the pixel group, if the ratio of the break to the total edge length is less than a predetermined value, the evaluation means means that the edge forming the peripheral edge of the pixel group forms the frame shape of the imprint. May be.

The proof image is a seal including a seal character, and the evaluation means is a process of extracting a line diagram from a pixel group consisting of color components extracted by the extraction process, an extracted line diagram, and a seal character. A process of matching with a print pattern may be performed, and a determination result of whether or not the pixel group constitutes a stamp may be made according to the result of the matching process.

The evaluation means includes a proof image based on whether the loaded double-sided originals include originals of different sizes and whether each of the plurality of double-sided originals is larger than the previous and next originals. The pair of two page data to be created may be specified.

The size setting of the plurality of originals is a mixed size, and in a plurality of double-sided originals to be read, a second size original smaller than the first size is inserted between two first size originals. If so, the evaluation means obtains page data corresponding to the first size 1 manuscript and page data corresponding to the first size other manuscript into two page data including a proof image. It may be a pair of.

The evaluation of whether or not the mirror image relationship is established by the evaluation means is
(1) A process of performing a mirror image conversion on a proof image included in one of the two page data.
(2) If the angle formed by the proof image after the mirror image conversion does not match the proof image contained in the other of the two page data, a process of performing rotation conversion on the proof image may be included.

The evaluation means further calculates the degree of deterioration of the diagram formed by the proof image in the two page data for each page data, and the page data having the larger degree of deterioration. However, the evaluation result that the proof image is included may be given.

The evaluation means may calculate the degree of deterioration from the line type, density, and interruption of the diagram constituting the proof image.

The image of the proof image is detected from each page data, and the image of the proof image is included in each of the pair of page data corresponding to the two document surfaces overlapping in the document bundle, and they have a mirror image relationship. It evaluates whether or not it is done, and based on the evaluation result, the display panel displays that the proof image has been transferred between the manuscripts. Therefore, if the user discards the transferred manuscript according to such display, the proof image will be displayed. It is possible to avoid external outflow. As a result, the risk of leakage of the proof image to the outside can be minimized.

It is a figure which shows the appearance of the document reading apparatus 1000. It is a figure which shows the internal structure of the document reading apparatus 1000. Document A to be converted into electronic data and scan data obtained by conversion to electronic data are shown. Shows the main routines of multiple flowcharts. It is a flowchart which shows the detail of the processing procedure of a specific area in a page. FIG. 6A shows a region near black and a region near vermilion in the RGB color space. FIG. 6B shows a color histogram of a black neighborhood region and a vermilion neighborhood region. The four edge groups 201, 202, 203, and 204 constituting the seal character and an example of the OCR database are shown. It is a flowchart which shows the detail of the usefulness determination procedure of a specific area. An example of the in-page information table 102T in the first operation example is shown. It is a flowchart which shows the detail of the determination procedure of the page pair to be compared. FIG. 11A shows an example of the device setting and the scanning setting of the document reading device 1000 in the first operation example. FIG. 11B shows an example of the page-to-page information table 104T in the first operation example. It is a flowchart which shows the detail of the originality determination of the comparison target included in a page pair. FIG. 13A shows an XY coordinate system for determining the mirror image relationship. FIG. 13B shows the process of rotational transformation. FIG. 14A shows an enlarged edge that is the target of the originality determination in the first operation example. FIG. 14B shows an in-page information table 102T in which the line thickness, density, and number of breaks are described in the first operation example. The notification screen displayed on the touch panel display 1001 in step S507 is shown. Indicates a bundle of originals consisting of mixed-size originals. The in-page information table 102T created for a bundle of manuscripts composed of mixed-size manuscripts is shown. FIG. 18A shows an example of the device setting and the scanning setting of the document reading device 1000 in the second operation example. FIG. 18B shows an example of the page-to-page information table 104T in the second operation example. FIG. 19A shows an enlarged edge that is the target of the originality determination in the second operation example. FIG. 19B shows an in-page information table 102T in which the line thickness, density, and number of breaks are described in the second operation example.

Hereinafter, embodiments of the document reader according to the present disclosure will be described.

[1] Appearance of Document Reading Device 1000 FIG. 1 is a diagram showing an appearance of the document reading device 1000. The document reading device 1000 is an MFP (Multi function Printer), and reads a bundle of documents placed on the tray 1004T of the ADF 1004 and converts it into electronic data according to a user operation performed on the touch panel display 1001. It is assumed that the imprint M1 on the front surface of the document D2 is transferred to the back surface of the document D1 and the transfer product M2 of the imprint is formed on the back surface of the document D1 due to the ink transfer in the document bundle. The document reading device 1000 according to the present embodiment detects a transferred material due to such ink reflection from the scan data obtained by reading the documents D1, D2, D3, D4, and D5, and displays the printed matter on the touch panel display 100 to the user. Notify.

[2] Internal Configuration of Document Reading Device 1000 FIG. 2 is a diagram showing an internal configuration of the document reading device 1000. As shown in this figure, the document reading device 1000 includes a document reading unit 101, an in-page identification unit 102, a comparison target effectiveness determination unit 103, an inter-page comparison target detection unit 104, a device setting unit 104S, and a page-to-page match determination unit. It is composed of 105 and a dialogue control unit 106.

(2-1) Document reading unit 101
When the document bundle to be read is placed on the tray of the ADF 1004 and the touch panel display 1001 is operated to start reading the document, the document reading unit 101 is placed on the tray 1004T shown in FIG. The double-sided documents are sequentially fed, and the front and back surfaces of each document are optically read. As a result, for example, scan data corresponding to odd-numbered pages and even-numbered pages on the front and back surfaces of the document can be obtained.

(2-2) Specific part 102 in the page
The in-page identification unit 102 calculates a color histogram for the in-page area of each page data, and in the color histogram, the distribution area of the color component of the pixel formed by the background or the color component of the pixel formed by the document body. A color having a frequency lower than the frequency of the distribution area of is detected as the color of the stamp, and the coordinates of the pixel group forming the color of the stamp are specified from the in-page area. In addition, the edge in the pixel group that forms the color of the stamp is extracted to try character recognition, and the presence or absence of the seal character in the pixel group that forms the color of the stamp is confirmed. In this way, if a pixel group that has the color of the seal and contains the seal character is detected, the pixel group is treated as forming a proof image, and a record indicating that the coordinates of the pixel group and the seal character are included. Is added to the in-page information table 102T. Here, the proof image is an image that proves that the normal image or the mirror image is responsible for the description of the manuscript, and in the present embodiment, not only the normal image of the imprint of the actual seal but also the mirror image of the imprint. Is also treated as a proof image.

(2-3) Comparison target effectiveness determination unit 103
The comparison target validity determination unit 103 extracts an edge from the peripheral edge of the area including the seal character specified by the in-page identification unit 102. When an edge with continuity is extracted, it is evaluated whether the edge is established as a figure. When it has the feasibility as a figure, the distinction between the figure shapes (including the distinction between the rectangular shape and the circular shape) is registered in the in-page information table 102T. Further, the brightness (L), saturation (S), and hue (H) of the pixels constituting the seal character are calculated and registered in the in-page information table 102T.

(2-4) Page-to-page comparison target detection unit 104
The page-to-page comparison target detection unit 104 sets whether or not the ADF1004 is attached in the image forming apparatus, whether or not the document reading setting is double-sided setting (double-sided / single-sided setting), and whether or not the size of the document to be read is mixed (mixed loading). / Same setting), the page pair to be compared is set in the page-to-page information table 104T. When reading originals of the same size on both sides, the set of consecutive pages in the image data is the page pair to be compared. It is a continuous page in the image data, and the combination of the 2.i (even) page and the 2.i + 1 (odd) page corresponds to the back surface of one manuscript and the front surface of the next manuscript. This is because they face each other when loaded.

When scanning a mixed-size double-sided document in which a smaller size document is inserted between two consecutive sheets of the same paper size, the two facing pages of the two sheets are displayed. The page pair to be compared is set in the page-to-page information table 104T. This is because if a large paper size document and a small paper size document overlap each other and there is an imprint containing stamp characters on the lower side of the document, the imprint can be transferred to the facing large size document. is there.

(2-5) Page-to-page match determination unit 105
The page-to-page match determination unit 105 compares the imprint contained in one page data of the acquired page pair with the imprint contained in the other page data of the page pair registered in the in-page information table 102T, and eventually Determine whether is the original or which is the transcript. In this comparison, the degree of deterioration of the degree of deterioration of the imprint on the page pair to be compared is calculated from the line width, density, and the number of breaks of the imprint contained on one side and many sides of the page pair to be compared. The imprint with a high degree of deterioration is regarded as a transcript, and the imprint with a low degree of deterioration is used as the original.

[3] First Operation Example (3-1) Page Data Obtained by Reading a Bundle of Documents With reference to the flowcharts of FIGS. 4, 5, 8, 10, and 12, the configuration is as described above. The operation of the document reading device 1000 will be described. In the first operation example, it is assumed that the document A shown in FIG. 3 is read and converted into electronic data. Document A is a document composed of three double-sided originals, and has an imprint on the surface of the second original.

When the ADF1004 was loaded, the papers were stacked without the stamps being dried, so that the imprint on the front surface of the second original was transferred to the back surface of the first original.

FIG. 4 is a flowchart of the main routine of the flowcharts of FIGS. 5, 8, 10, and 12. In step S101 of FIG. 4, the document bundle is read by the ADF1004 to obtain scan data. The scan data A in FIG. 3 is data obtained by optically reading a bundle of documents of the document A, and has a page structure in which the front surface of each document constituting the document A is an odd number of pages and the back surface is an even number of pages. .. Therefore, on the second page of the scan data, there is an imprint image due to the transfer of the undried ink.

(3-2) Color Histogram Analysis When the scan data is obtained, the in-page identification unit 102 starts processing. The in-page identification unit 102 identifies an area existing on each page of the scan data and obtains the in-page information table 102T (step S102). The procedure of step S102 can be expanded to the subroutine of FIG. FIG. 5 is a flowchart showing the details of the processing procedure for specifying the in-page area.

The variable i in this flowchart is a variable that indicates each page of scan data, and in this flowchart, serves as a control variable that controls the repetition by the loop of steps S201 to S212. First, the variable i is initialized (step S201), and the color histogram of page i is calculated (step S202). The color histogram generated in this way shows the highest frequency among the main color components of the document, and determines whether the frequency is lower than the next rank with a color component different from the main color components of the base of the document and the text. (Step S203).

Manuscripts such as slips are mainly composed of characters, ruled lines, etc., and the pixels corresponding to these characters and ruled lines form an area centered on black and gray (black neighborhood area 200B) in the RGB color space of FIG. 6 (a). Occupy. The pixels corresponding to the imprint occupy different regions (specific color region 200S) in the RGB color space of FIG. 6A.

The black vicinity region 200B shows the highest frequency in the color histogram of FIG. 6 (b). The specific color region 200S shows the next highest frequency after the black neighborhood region 200B in the color histogram of FIG. 6 (b).

When a part of the pixel group of the page data is composed of pixels having a color component different from the main color component of the document and the pixel shows a frequency of the next rank or lower, a record of the pixel group showing the power of the next rank or lower. Is added to the in-page information table 102T (step S204). Hereinafter, the pixel group added as a record is simply referred to as a pixel group. Further, the in-page identification unit 102 has the coordinates (X1, Y1), (X4, Y4), vertical widths H1, H2, and horizontal widths of the imprints (X1, Y1), (X4, Y4), the second page of the paper, and the third page of the paper in the record of the in-page information table 102T. Register W1 and W2.

(3-3) Edge extraction / character recognition Since the page identification processing has been completed, the edge processing is started. First, the inside of the pixel group on page i is divided by a blank background to detect the edge group (step S205), and it is determined whether or not the edge has continuity of a certain level or more (step S206). When an edge having continuity exists on page i (Yes in step S206), the edge of the pixel group is extracted as a line drawing (step S207).

Subsequently, the feasibility of the seal character is determined. It is determined whether the matching rate between the edge of page i and the type pattern is below the threshold value (step S208). When the value is below the threshold value (Yes in step S208), the edge having continuity is recognized as the seal character (step S209), and the corresponding record of the in-page information table 102T indicating that the edge in the pixel group constitutes the seal character. Register in the image / feature column of (step S210).

As shown in FIG. 7, the imprint of an example of FIG. 3 has four edge groups 201, 202, 203, and 204 with the horizontal and vertical base areas 102B and 102C as boundaries, and thus such edge groups are extracted. Then, the matching rate with the OCR type pattern is calculated, and character recognition of each edge group is attempted.

The rough relationship between the characters to be recognized and the matching rate will be described below. Correspondence table 102L of FIG. 7 shows the matching rate from 1 to 5 and the edge group corresponding to each matching rate. The edge group in the correspondence table 102L constitutes a type such as "experiment", "character", and "ma".

As shown in this figure, edges that are close to print are recognized as having a high matching rate.

Therefore, the importance level of the seal character is determined from the matching rate, and as shown in FIG. 9, the importance level is set in the image / feature column of the records corresponding to the second page and the third page of the in-page information table 102T. To register. In the present embodiment, if the matching rate of characters formed by the edges of the pixel group is 2 or less, the importance level is set to "high", and if the matching rate of characters formed by the edges of the pixel group is 3, it is important. Set the degree level to "medium". If it is 4 or more, the importance level is set to "low".

In FIG. 7, the imprint characters on the second and third pages of the scan data are in a seal script, and the matching rate with the OCR type pattern is calculated to be 1 or 2 mag lower. Therefore, in the in-page identification unit 102, the pixel groups of the second and third pages form the seal characters in the image feature columns C42 and C43 of the records of the second page and the third page of the manuscript in FIG. It is registered that the importance level is high (step S210). Step S211 is a step for determining whether or not the loop continuation requirement of steps S202 to S212 is satisfied. If the variable i is less than the maximum number of pages (Yes in step S211), the variable i is incremented (step S212). , Return to step S202.

Hereinafter, as long as the variable i falls below the page maximum value, the repetition of steps S202 to S212 continues. When the variable i reaches the maximum number of pages and the processing of all pages is completed, step S211 defining the continuation requirement becomes No. When step S211 becomes No, the process of this flowchart is terminated, and the process returns to the main routine of FIG.

(3-4) Extraction of Graphic Frame Since the in-page identification unit 102 has finished processing, the comparison target validity determination unit 103 starts processing in step S103, and the pixel group specified by the in-page identification unit 102 Determine if it is useful as a seal. Specifically, it is determined whether or not the continuous edges have the feasibility as a figure. The procedure of step S103 described above can be expanded to the subroutine of FIG.

FIG. 8 is a flowchart showing the details of the procedure for determining the usefulness of the specific area. Similar to FIG. 5, the flowchart of FIG. 8 also uses the variable i that identifies each page as a control variable. In the flowchart of FIG. 8, the variable i is set to 1 (step S301), and it is determined whether or not an imprint (referred to as a comparison target) to be compared between pages exists on the page i (step S302). When the comparison target exists (Yes in step S302), the edge at the peripheral edge of the pixel group specified on page i is extracted (step S303). When the edges are extracted in this way, it is determined whether or not the detected edges have a certain degree of continuity or more (step S304), and whether or not the detected edges are established as a figure (step S305). The frame shape of the seal includes a circular shape and a rectangular shape. The in-page identification unit 102 extracts these circular and rectangular edges from the periphery of the pixel group. Even if there is a break in the edge extracted from the peripheral edge of the pixel group, if the ratio of the break to the total edge length is less than or equal to a predetermined value (for example, 10% of the total length), the edge has continuity and is a seal. It shall form the frame shape of. If both steps S304 and S305 are Yes, the area specified on page i is extracted as the area of the seal, and as shown in FIG. 9, the shape of the record in the in-page information table 102T / upper left coordinate column C22, Register in C23 (step S306). If step S304 is Yes and S305 is No, extraction is not performed from the area specified on page i, and the process proceeds to step S311 to increment the variable i. In this way, after extracting the seal, the coordinates of the extracted region in the HCl color space are extracted.

Since there is continuity at the edge of the periphery of the imprint on the second and third pages, the frame of the seal is established as a figure. Therefore, the comparison target validity determination unit 103 tells the record of the imprint on the second and third pages of the in-page information table 102T that the shape is rectangular, and that the width and height of the rectangular shape are rectangular. sign up.

If step S304 is No, steps S305 to S308 are skipped. Step S309 is a determination as to whether or not the loop continuation requirement is satisfied. If the variable i is less than the maximum number of pages (Yes in step 309), the variable i is incremented (step S310), and the process returns to step S302. As long as the variable i falls below the page maximum, the step repeats. As a result, the individual pages are subjected to processing. When the variable i reaches the maximum number of pages and the processing of all pages is completed, step S309 for defining the end requirement becomes No, the processing of this flowchart is terminated, and the process returns to the main routine. Due to the return from the subroutine, the process is restarted from step S104 immediately after step S103.

(3-5) Identification of Page Data Pairs to be Compared In step S104, the page pair to be compared is determined according to the presence / absence of ADF attachment and the original size, and the page-to-page information table 104T is created. The procedure of step S104 can be expanded to the subroutine of FIG. FIG. 10 is a flowchart showing the details of the procedure for determining the page pairs to be compared.

Step S401 of this flowchart is for determining whether or not the ADF1004 is attached, and step S402 is for determining whether or not double-sided reading is set. If any of these steps S401 and S402 is No, no processing is performed and the processing of this flowchart ends.

If both steps S401 and S402 are Yes, it is determined whether or not the reading settings of the document reading device 1000 are mixed sizes. If the documents to be read have the same size (No in step S403), the process proceeds to the loop consisting of steps 404 to S408. In the loop of steps S404 to S408, the variable i is set to 1 (step S404), it is determined whether or not pages 2 and i include the comparison target (step S405), and if so, the 2nd and i-th pages are included. 2, i + 1 The first page is the page to be compared vs. u, v (step S406).

The device settings and scanning settings of the document scanning device 1000 are the settings shown in FIG. 11A, that is, the ADF1004 is mounted, the single-sided / double-sided setting is set to double-sided, and the document size is set to the same page. And. Since the device settings are double-sided scanning and the document size is the same, the page pairs to be compared are 2 · i pages and 2 · i + 1 pages (step S406).

As a result, as shown in an example of the inter-page information table 104T in FIG. 11B, the other page is defined as "3" in the two pages of the scan data. Also, on the third page of the scan data, the other page is defined as "2". Since imprints are present on the second and third pages of the scan data, as shown in FIG. 11B, the records on the second and third pages of the in-page information table 102T have a comparison target of "Yes". , And the necessity of judgment is specified in "Implementation required".

Step S407 is a loop continuation requirement, and if the page number of the second i + second page is lower than the last page (Yes in step S407), the variable i is incremented (step S408), and the process returns to step S405. As long as the continuation requirement of step S407 is satisfied, the above series of processes are executed for the information elements of the next rank. 2. When the page number of the i + 2nd page reaches the last page, this subroutine is terminated and the process returns to the main routine. By the return from the subroutine, the process is restarted from the step S105 immediately after.

In step S105, the originality of each comparison target included in the page pair to be compared is determined. The procedure of step S105 can be expanded to the subroutine of FIG. FIG. 12 is a flowchart showing details of the originality determination of the comparison target included in the page data pair.

The variables u and v in this flowchart are variables that indicate the set of pages that should be the target of the match determination. First, page u is the first page having a comparison target in the page-to-page information table 104T (step S501). Further, in the page-to-page information table 104T, the page on the other side of the page u is set as the page v (step S502).

(3-6) Determination of Mirror Image Relationship Next, it is evaluated whether or not the comparison target of page u and the comparison target of page v have a mirror image relationship (step S503).

Judgment as to whether or not the mirror image relationship is established is made by going through the following process. The comparison target of the page data on the second page is the comparison target of the page data on the third page in the first quadrant of the XY coordinate system as shown in FIG. 13 (a) (the X coordinate is positive and the Y coordinate is positive). Arrange in the second quadrant (quadrant where the X coordinate is negative and the Y coordinate is positive).

When arranging the pixel group that is the imprint to be compared, the upper right coordinates (X ₁ ', Y _{1 ') of the imprint on page 2 and the upper left coordinates (X 4} , Y ₄ ) of the imprint on page 3 are Y. The offset of the upper left coordinate of the imprint of page 2 (ofset1, ofset3) and the offset of the upper left coordinate of the imprint of page 3 (ofset2, ofset4) are determined so as to be line-symmetrical on the axis. This is because the positions of the pixel groups in the pair of page data may be shifted vertically and horizontally. In this way, when the comparison target on the second page and the comparison target on the third page are arranged in the second quadrant and the first quadrant, the comparison target on the second page and the comparison target on the third page are in a relationship of folding back on the Y axis. Therefore, the first-order transformation matrix A that wraps around the Y-axis attempts to transform the individual pixels that make up the imprint of the second quadrant.

The first-order transformation matrix A that wraps around the Y-axis is shown in Equation 1 below.

When the imprint of the second page is converted by the linear transformation matrix A and matches the imprint of the third page, the judgment result that the imprint of the second page and the imprint of the third page are in a mirror image relationship is determined. Give it down.

In the transformation by the linear transformation matrix A, when both imprints do not match, the imprints are transformed by the rotation matrix as shown in FIG. 13B, and the imprints on the second page are rotated by the minute angle Δθ. The rotation of Δθ is performed, and it is determined whether or not the imprint after conversion, which is converted by the linear transformation matrix A, matches the imprint on the third page. If they match, it is determined that both are in a mirror image relationship. To give. When the imprint after the rotation conversion of the minute angle Δθ does not match the imprint of the third page, the rotation conversion of the minute angle Δθ and the determination of whether or not it matches the imprint of the third page are repeated.

(3-7) Judgment of Deterioration Degree When the comparison target in the page data pair does not match even if the rotation of the comparison target is repeated (No in step S503), the comparison target of page u and the comparison target of page v are Assuming that it is not related to the mirror image, the processing of this subroutine is terminated and the process returns to the main routine.

Through the above process, when the angles of the comparison targets in the page data pair match (Yes in step S503), it is assumed that the comparison target of the page u and the comparison target of the page v are in a mirror image relationship. Subsequently, the line width, density, and break at the edge of the pixel group in pages u and v are acquired (step S504), and the degree of deterioration of pages u and v is calculated from the acquired line width, density, and break (step). S505). Subsequently, the magnitude of the deterioration degree v of the page u and the deterioration degree v of the page v are compared.

Here, as shown in the enlarged view of FIG. 14A, the edge of the imprint of the second page has a narrow line width (the line width of the second page is 2, while the line of the third page). The width is 5), and the edge of the second page has a low density (the edge E1 of the second page has a density of 50, and the edge E2 of the third page has a density of 75). Further, it is assumed that the second page has more breaks (five breaks) than the third page. The page-to-page match determination unit 105 acquires these line widths, densities, and the number of breaks, and registers them in the records on the second and third pages of the in-page information table 102T (step S504). Then, from these line widths, densities, and breaks, the degree of deterioration of the comparison target on pages u and v is calculated (step S505), and whether or not the degree of deterioration of page u and the degree of deterioration of page v match. If it is determined (step S506) and the degree of deterioration of page u and the degree of deterioration of page v do not match (No in step S506), the transcript exists on the page u and v on the side with the larger degree of deterioration. Notify the user to do so (step S507).

The imprint on page 3 is calculated to have a low degree of deterioration because it has a high density, a thick line width, and few breaks. On the contrary, the imprint of page 2 is calculated with a high degree of deterioration. Therefore, it is determined that the degree of deterioration of the second page and the degree of deterioration of the third page do not match, and the user is notified that the transcript exists on the second page having a large degree of deterioration (step S507).

FIG. 15 shows a notification screen displayed on the touch panel display 1001 in step S507. The notification screen 1001D of FIG. 15 indicates that the imprint of the second front surface has been transferred to the back surface of the first sheet of the scanned document, and it is necessary to recreate the one-page document and perform rescanning. Show the user that there is.

Step S510 of FIG. 12 is a determination of whether or not the loop continuation requirement is satisfied, and determines whether or not a page including a comparison target exists after the page v. If it exists (Yes in step S510), in the page-to-page information table 104T, the page having the comparison target, which is next to the page v, is set to the page u (step S511), and the process returns to step S502.

As long as there is a page containing the comparison target after the variable v, the repetition of steps S502 to S507 and S510 to S511 continues. As a result, the individual pages are subjected to processing. If there is no page including the comparison target after the variable v, the processing of all pages is completed, step S510 for defining the end requirement becomes No, the processing of this flowchart is ended, and the process returns to the main routine. .. The return from the subroutine also ends the processing of the main routine.

[4] Second Operation Example (4-1) Document Bundles Read in Mixed Mode In the second operation example, a document bundle of mixed loading size as shown in FIG. 16 is read. In this document bundle, a small-sized document B is inserted between the normal-sized documents A and C. An imprint is present on the surface of the first original, and an imprint is present on the surface of the third original. By reading the mixed document bundle in the double-sided mode using the ADF1004, scan data as shown in the lower part of FIG. 16 can be obtained.

This scan data consists of six page data, and each page data corresponds to the front surface and the back surface of each document in the mixed document bundle. Among the scan data, the page data of the first page, the second page, and the fifth page contains an imprint image obtained by reading the imprint of the first original and the imprint of the third original. The imprints on pages 1, 2 and 5 of this scan data are vermilion and red, and appear in a portion of the RGB color space that is far from the color distribution of the original body. Therefore, the in-page identification unit 102 detects the upper left coordinates of the imprint portion having a peculiar color distribution from pages 1, 2 and 5, and records the records including the detected upper left coordinates as shown in FIG. Add to table 102T.

Since the imprint on the surface of page 1 has a rectangular shape and the imprint on pages 2 and 5 has a circular shape, as shown in FIG. 17, a record showing the width / height (W1 × H1) and the upper left coordinate of the rectangular shape on the first page. , A record showing the diameter (φ1) and center coordinates (X3, Y3) of the circular shape on the second page, and the diameter (φ1) and center coordinates (X4, Y4) of the circular shape on the fifth page in the in-page information table. Register in the 102T record.

The device settings and read settings are as shown in FIG. 18A. The ADF1004 is already mounted, the single-sided / double-sided setting is set to double-sided, and the same size / mixed loading is set to mixed loading. Further, by detecting the paper size at the time of reading the ADF, it is detected that the first and third sheets of the original are the normal size and the second sheet is the small size. Since the mixed loading mode is set, the size of the paper is registered in the corresponding record of the interpage information table 104T each time the originals are fed out from the document bundle placed on the tray 1004T of the ADF1004. The processing of the flowchart of FIG. 10 is executed according to the device setting, the reading setting, and the paper size detection.

(4-2) Determining the page data pair to be compared In the second operation example, since the ADF1004 is mounted, the scanning mode is set to double-sided scanning, and the document sizes are mixed, step S401 is Yes, and step S401 is a step. S402 becomes Yes, step S403 becomes Yes, and the process proceeds to step S411. Steps S411 to S418 define a loop with the variable i as a control variable. Step S411 is the initialization of the variable i, and 1 is assigned to the variable i. Step S412 is a determination as to whether or not page 2 and i include a comparison target, and step S413 is a determination as to whether or not the size of page 2 and i exceeds the size of page 2 and i + 1. When the size of page 2 · i exceeds the size of page 2 · i + 1 (Yes in step S413), the pair of page data to be compared is determined in consideration of the mixed size of the document bundles.

When a bundle of documents is a mixed size and a large size document and a small size document overlap and face each other, the page to be transferred changes depending on where the imprint is located on the large size document. ..

When the imprint exists in the area of the large-sized document that overlaps with the small-sized document, page 2i and page 2i + 1 are page pairs as in the first operation example.

If an imprint exists in an area of a large-sized document that does not overlap with the small-sized document, page 2i and page 2i + 2m + 1 form a page pair. In the second operation example, the page pairs to be compared are determined in consideration of these cases. That is, in step S416 of FIG. 10, the combination with page 2i and page 2i + 1 and the combination with page 2i and page 2i + 2m + 1 are set as page pairs to be compared.

In the following step S417, the step is a determination as to whether or not the loop continuation requirement is satisfied, and when pages 2, i + 2, m + 2 are less than the number of final pages (Yes in step S417), the variable i is incremented (step S418). ), Return to step S412.

When the size of page 2 · i is equal to or smaller than the size of page 2 · i + 1 (No in step S413), page 2 · i and page 2 · i + 1 are set as page vs. u, v to be compared (No). Step S414). Step S417 is a determination as to whether or not the loop continuation requirement is satisfied. If 2.i + 2 is less than the number of final pages (Yes in step S415), the variable i is incremented (step S418), and the process returns to step S412. .. As long as the variable i falls below the page maximum value, the repetition of steps S421 to S418 continues. As a result, the individual pages are subjected to processing. When any of steps S415 and S417 becomes No, it is assumed that the processing of all pages is completed, the processing of this subroutine is terminated, and the process returns to the main routine.

The bundle of documents read in the second operation example is a normal size document on the first and third sheets, a small size document on the second sheet, and the first and second pages of the scan data are normal size pages. The data, the 3rd and 4th pages are the small size page data, and the 5th and 6th pages are the normal size page data. Since there is only one small-sized document, m = 1.

Therefore, when i = 1, page 2 (= 2 · i) is set as page u, and page 5 (= 2 · i + 2 · m + 1) is set as page v (step S416). As a result, as shown in 141 and 142 of FIG. 18B, pages 2 and 5 are set as page pairs to be compared and are described in the page-to-page information table 104T.

Since imprints exist on the 2nd and 5th pages of the scan data, the records on the 2nd and 5th pages of the in-page information table 102T are set as "Yes" for comparison, and the necessity of judgment is determined. It is stipulated in "Implementation requirements". When the comparison target and the necessity of determination are set, the page-to-page match determination unit 105 starts processing.

As shown in the enlarged view of FIG. 19A, the edge E11 of the imprint of page 2 has a narrow line width (the line width of the second page is 2, while the line width of the fifth page is 5). Also, the edge E12 of the imprint on the second page has a low density (the edge on the second page has a density of 50, and the edge on the fifth page has a density of 80). Further, since the second page has more breaks than the fifth page, the page-to-page match determination unit 105 provides information on the line width, density, and number of breaks in the page as shown in FIG. 19B. Register in table 102T.

Since the imprint of page 5 has a high density, a thick line width, and few breaks, the imprint of page 5 is the original and the imprint of page 2 is the transcript in determining the originality. Then, that fact is registered in the corresponding record of the in-page information table 102T.

[5] Summary As described above, according to the present embodiment, the portion where the seal and the signature are made is specified from the page area of the scan data by performing color histogram analysis and character recognition on the internal area of the page. Then, since it is determined whether these parts are included in the page pair to be compared and which of the imprints of these parts has a mirror image relationship and which is deteriorated, a plurality of manuscripts are used in the ADF1004. Even when the stamps that are the original and the stamps that are the transferred products are misaligned, the presence or absence of ink transfer can be determined with appropriate determination accuracy.

[6] Modifications Although the description has been made based on the embodiment of the document reading device according to the present disclosure, it goes without saying that the present disclosure is not limited to the above-described embodiment, and the following modifications can be considered.

(1) The image of the imprint has been described as an example as a proof image that proves that the person is responsible for the description on the manuscript paper. However, it is not limited to this. Any other image may be used as the proof image as long as it is an image that proves that the person is responsible for the description of the manuscript. For example, an image of a user's handwritten signature may be used as a proof image. Further, the image obtained by imprinting the fingerprint may be used as the proof image. Since ink is used for these signatures and seals, there is a possibility that transfer due to ink transfer may occur.

(2) When the specific unit 102 in the page recognizes characters, supplementary processing of unrecognized characters may be executed. Specifically, when there is an unrecognizable edge in the scan data page, the page identification unit 102 performs mirror image conversion on the edge in the scan data page, and the character after the mirror image conversion is recognized as a character. You may try to re-recognize whether or not. When character recognition becomes possible by performing mirror image conversion, the fact that the character is a transcript is registered in the in-page information table 102T. If the determination accompanied by such mirror image conversion is performed, it becomes unnecessary to determine whether or not the transcript is a transcript by referring to the line width, density, breakage, etc., and the processing can be simplified.

(3) In determining whether or not the mirror image relationship is established, the rotation of a minute angle is repeated, but the present invention is not limited to this. If the imprint contained in the pair of page data has a rectangular frame, or if there is a background boundary between the stamp characters, the rotation of the transferred image is performed from the angle formed by the sides of these rectangular frames and the background boundary. The angle may be calculated and the transferred image may be rotated by this rotation angle.

(4) In edge extraction, feature points may be extracted from each pixel group of a pair of page data, and the same or not of the transferred image after mirror image conversion may be determined by comparing the feature points.

(5) An OCR database of character shape patterns is set up for each of the special fonts (seal script, seal script, etc.) that are difficult to forge and are often used for stamps, and the matching rate is used in the OCR database of any character shape pattern. May be determined whether or not becomes higher.

(6) In the above embodiment, in the color histogram, the specific portion 102 in the page has a frequency lower than either the distribution area of the color component of the pixel formed by the background or the distribution area of the color component of the pixel formed by the document body. The pixel of the color component is the color of the stamp, but it is not limited to this. If the color components of the pigments that make up a typical stamp, such as the color of red meat, are registered in advance, and if such a color component and a pixel group of a color in the vicinity thereof exist inside the page data, the specific portion 102 in the page will be displayed. , Such a pixel group may be specified as an imprint area. The neighboring color here means a color in which the spatial distance from the vermilion color in the RGB color space is less than a predetermined distance.

(7) In the above embodiment, when the transfer of the imprint is found, the notification screen is displayed on the touch panel display 1001, but the present invention is not limited to this. The user may be notified of the occurrence of transfer due to ink transfer by speaking a voice narration that reads out the document in which the transferred image exists. In addition, the blinking of the LED indicator may notify the user of the occurrence of transfer due to ink transfer. By transmitting message data to a device existing on the premises, the user may be notified of the occurrence of transfer due to ink transfer.

(8) Among the large-sized documents, the page pair to be compared may be determined depending on whether or not the imprint exists in the area overlapping the small-sized document. If the imprint exists in the area of the large-sized document that overlaps with the small-sized document, page 2i and page 2i + 1 are paired with each other. Further, when an imprint exists in an area of a large-sized document that does not overlap with the small-sized document, page 2i and page 2i + 2m + 1 are page pairs.

(9) The document reading device 1000 of the above embodiment is assumed to be an MFP, but the present invention is not limited to this. The document reading device 1000 may be a single-function peripheral device (scanner) of a personal computer.

INDUSTRIAL APPLICABILITY According to the present invention, it can be easily detected from one of two facing double-sided originals that a proof image is placed on the other from each page data obtained by reading the loaded double-sided originals. It may be used in various industrial fields such as the information equipment industry, retail, leasing, real estate, advertising, transportation, and publishing.

101 Manuscript reading unit 102 Page-specific information table 102T Page-to-page information table 103 Comparison target effectiveness judgment unit 104 Page-to-page comparison target detection unit 104S Device setting unit 104T Page-to-page information table 105 Page-to-page match judgment unit 106 Dialogue control unit 1000 Device 1001 Touch panel display 1004 ADF
1004T tray

Claims (11)

It is a document reading device that reads a plurality of loaded double-sided documents and obtains page data corresponding to each of the double-sided sides of each document.
One of the two page data corresponding to the two manuscript surfaces facing each other in the state of the manuscript bundle contains a proof image that proves that the person is responsible for the description of the manuscript, and the other is the proof image. An evaluation means for evaluating whether or not an image having a mirror image relationship is included in the image,
A document reading device including a notification means for notifying that a proof image has been transferred between documents based on an evaluation result. The evaluation means
Processing to calculate the color histogram of each page data,
An extraction process is performed to extract the distribution area of the color components having a frequency lower than the distribution area of the main color components of the manuscript body from the color histogram.
The document reading apparatus according to claim 1, wherein a pixel group composed of color components in a distribution area extracted by the extraction process is used as a proof image. The proof image is an image of an imprint formed by imprinting a manuscript.
The evaluation means
The document reading apparatus according to claim 1, wherein a proof image is obtained by extracting a pixel group consisting of a color component of a pigment used for imprinting and / or a color component of a color in the vicinity thereof. The proof image is an image of an imprint formed by imprinting a manuscript.
The document reading device according to claim 2, wherein the evaluation means performs a process of extracting the frame shape of the imprint. Even if there is a break in the edge forming the peripheral edge of the pixel group, if the ratio of the break to the total edge length is less than a predetermined value, the evaluation means means that the edge forming the peripheral edge of the pixel group forms the frame shape of the imprint. The document reading device according to claim 4, wherein the document reading device is characterized by the above. The proof image is a seal containing the seal characters, and is
The evaluation means
A process of extracting a diagram from a pixel group consisting of color components extracted by the extraction process,
Performs processing to match the extracted diagram with the type pattern of the seal character,
The document reading apparatus according to any one of claims 2, 4, and 5, wherein a determination result of whether or not the pixel group constitutes a seal is made according to the result of the matching process. The evaluation means includes a proof image based on whether the loaded double-sided originals include originals of different sizes and whether each of the plurality of double-sided originals is larger than the previous and next originals. The document reading apparatus according to any one of claims 1 to 6, wherein the pair of two page data is specified. The size setting of the plurality of originals is a mixed size, and in a plurality of double-sided originals to be read, a second size original smaller than the first size is inserted between two first size originals. If so, the evaluation means obtains page data corresponding to the first size 1 manuscript and page data corresponding to the first size other manuscript into two page data including a proof image. The document reading device according to claim 7, wherein the document reading device is paired with the above. The evaluation of whether or not the mirror image relationship is established by the evaluation means is
(1) A process of performing a mirror image conversion on a proof image included in one of the two page data.
(2) When the angle formed by the proof image after the mirror image conversion does not match the proof image contained in the other of the two page data, the process of performing rotation conversion on the proof image is included.
The document reading device according to any one of claims 1 to 8. The evaluation means further calculates the degree of deterioration of the diagram formed by the proof image in the two page data for each page data.
The document reading apparatus according to any one of claims 1 to 9, wherein the page data having a larger degree of deterioration gives an evaluation result that the image for proof is included. The document reading device according to claim 10, wherein the evaluation means calculates the degree of deterioration from the line type, density, and interruption of the diagram constituting the proof image.

Images