JP4783317B2 - Image forming apparatus, image processing method, program for causing computer to execute the method, and recording medium - Google Patents

Description

  The present invention provides an image processing apparatus, an image processing method, and a computer-implemented image processing apparatus having page number information adding means and page rearranging means that are free from damage due to fixing and do not impair the degree of freedom of user document creation. And a recording medium.

In recent offices, not only paper documents are copied, but also the spread of PCs and the enhancement of network infrastructure, making it easy to digitize by scanning the manuscript output as paper, and easily via the network It has been frequently said that it is useful for knowledge creation by managing and sharing resources with others.
For example, the materials of the presentations that you participated in are scanned and viewed from outside, or shared internally. Scanning is generally performed by setting each paper document on a flatbed scanner and pressing the scan start button. However, if ADF (Auto Document Feader) is used, the stacked paper is automatically It is very convenient because it is read.
Collecting a plurality of scanned originals into a single file, such as a multi-page Tif, in page order, eliminates the need to operate on multiple files in order to view a single document. is there.

On the other hand, when using ADF, misfeeds may occur in rare cases. There are various reasons for this, but when a misfeed occurs, the page is naturally omitted, and sometimes the area where the important information is recorded may be lost. Even if a misfeed is detected and a page that has been misfeeded is scanned later, the order of the pages arranged in one file will be out of order. In this case, the viewability of the electronic document obtained by scanning becomes very low.
Furthermore, it is very troublesome to place paper originals in page order and set them in the ADF. This is not limited to scanning, but of course there is a similar problem when copying and outputting.

In order to avoid such a problem, conventionally, RFID (Radio Frequency Identification) with page information is added to the paper to be output, and if an abnormality is found in the order of pages, a warning is issued. In order to arrange them in the correct order, an idea of automatically rearranging them has been made (Patent Documents 1 and 2). However, considering the application of the above-described conventional technology to electrophotography, it is conceivable that a large pressure and heat are applied during fixing and the RFID is damaged. An electrophotographic output method will be described later in Examples.
In another conventional technique, an invention has been devised in which a bar code with page information is added to a specific area of a document to issue a warning or arrange the page order as in the above invention (Patent Document 3). ). However, according to this method, there is a problem that the user cannot write in a specific area and the user's degree of freedom in document creation is impaired.

Furthermore, as information adding means that does not impair the degree of freedom of user document creation, for example, there is a technique called digital watermark (Patent Document 4). The digital watermark is a means for adding information to an image so as not to be seen by human eyes. For example, it can be realized by embedding a specific pattern in a high-frequency region that is relatively difficult for humans to recognize. However, when the amount of information to be added increases, the image quality deteriorates significantly, and it can be said that the image quality and the amount of additional information are in a trade-off relationship.
Furthermore, in general, a digital watermark is often used for improving security, and for example, a user name, the number of copies, and the like are often added. As a result, when information that is not recognized by the digital watermark is added, the area that can be used as the page number is often very limited. For example, assuming that 8 bytes of information are added and 6 alphabets are used as a user name, the amount of information required for 1 alphabet is 1 byte, so only 2 bytes can be used as a page number. The page number information added to different document data is continuous with a probability of at least 1/65535 (2 bytes can represent numbers up to 65535).

More specifically, if page numbers are added according to similar rules in different printers, the probability is even higher. For example, when page number information is added in order from 1 in each printer, this is clear.
In Patent Document 5, when the original surface of the first page is read, the CPU determines whether or not the page is blank based on the ratio of white data in the image data of the first page, and the first page is blank. If the first page is not reversed, the image data in the main storage unit is rearranged in the reverse order and stored in the storage unit or printed out to the printer unit. Is determined to have an image only on one side or on both sides, and if it is a single-sided original, it is stored or printed as correctly set, and if it is a double-sided original, the first page is displayed on the display unit 5. A technique is disclosed in which a user is allowed to confirm and save or print if correct, or rearrange and save or print if different.
JP 2006-254058 A JP 2006-130784 A JP 2005-153326 A Japanese Patent No. 3570213 JP 2001-251383 A Koichi Kise and Osamu Yanagida, “Area Image Segmentation Using Thinning of White Areas”, IEICE Transactions, No. 6, pp. 1608-1616, 1997 C. Cortes and VN Vapnik, “Support vector Networks,” Machine Learning, vol.20, pp.273-297, 1995

That is, the prior art has the following problems.
First of all, when a misfeed occurs, there is a problem that a page is omitted as a matter of course, and sometimes an area in which important information is recorded is lost.
In addition, the conventional technology that adds RFID with page information and issues a warning when an abnormality is seen in the page order, or automatically rearranges whatever order they are in order. Considering application to photography, there is a possibility that RFID is damaged due to large pressure and heat applied during fixing.
In addition, by adding a bar code with page information to a specific area of the document and issuing a warning or arranging the page order, the user cannot write to the specific area, creating a user's document. There was a problem that said it would impair the degree of freedom.
In addition, in the method of embedding a digital watermark, which is a pattern peculiar to the high-frequency region that is relatively difficult for humans to recognize, as the amount of information to be added increases, the image quality deteriorates significantly, and the image quality and the amount of additional information trade off. There was a problem with the relationship.

The present invention has been made in view of the above-described conventional problems, and an object of the present invention is to provide page number information adding means and page rearrangement that do not cause damage due to fixing and do not impair the degree of freedom of user document creation. An image forming apparatus provided with a means, an image processing method, a program for causing a computer to execute the method, and a recording medium are provided.
Another object of the present invention is to provide an image forming apparatus provided with page rearrangement means capable of rearranging pages accurately even with output images obtained from different printers while minimizing degradation of image quality. An image processing method, a program for causing a computer to execute the method, and a recording medium are provided.

In order to achieve the above-described object, the invention according to claim 1 is an image forming apparatus that reads a plurality of pages of a document to form an image, and that reads the document and generates image data; Information embedding means for embedding information having at least the page number of the original document in a digital watermark format in the original read by the image reading means; information extracting means for extracting information embedded in the original; and the information extracting means Page number acquisition means for acquiring the page number of the document from the read information, storage means for storing image data of a plurality of pages read by the image reading means as one group, and the page acquired by the page number acquisition means Page number proximity determining means for determining whether the number and each page number already stored in the storage means are close numbers; When the page number proximity determining means determines that there is a close page, the image similarity determining means for determining the image similarity of the image data of the plurality of pages, and the image determined by the image similarity determining means The apparatus further comprises a page order determining unit that includes a page rearranging unit that rearranges the pages of the original according to the similarity and group-integrates the pages of the original .

Also, an invention according to claim 2, wherein the page order determination unit, rearrange the pages of the document, or when performing integration of a group of pages of the document, be provided with a confirmation means for user confirmation It is characterized by.
According to a third aspect of the present invention, the image similarity determination unit includes a document type identification unit that determines at least a document type of a page of the document.
According to a fourth aspect of the present invention, when the page number acquired by the page number acquisition unit is 0, the top page number of image data composed of a plurality of pages read by the image reading unit is randomly determined. And embedded information generating means .

Also, fifth aspect of the present invention, the steps of generating an image processing method in an image forming apparatus for forming an image by reading an original of a plurality of pages, the image data by reading the original document by the image reading means, A step of embedding information having at least the page number of the manuscript in an electronic watermark method in the manuscript read by the image reading unit by the information embedding unit; a step of extracting information embedded in the manuscript by the information extracting unit; A step of acquiring a page number of the original from information read by the information extraction unit by a number acquisition unit, a step of storing a plurality of pages of image data read by the image reading unit by a storage unit as one group, and a page The page number acquired by the page number acquisition means by the number proximity determination means and the existing number A step of determining whether each page number stored in the storage means is a close number; and when the page number proximity determination means determines that there is a close page, the image similarity determination means determines the plurality of the page numbers. The step of determining the image similarity of the image data of the page, and the page rearrangement unit rearrange the pages of the original according to the image similarity determined by the image similarity determination unit, and the page of the original And a step of group integration .

According to a sixth aspect of the present invention, there is provided a program for causing a computer to execute the image processing method according to the fifth aspect.
According to a seventh aspect of the invention, there is provided an information recording medium on which a computer program for achieving the image processing method according to the fifth aspect is described .

According to the present invention, it is possible to embed page number information in an image, extract it at the time of reading, determine the page order of the document, and rearrange the pages of the document based on the determined page order. When a feed or the like occurs, an appropriate process can be presented to the user, and the convenience for the user is improved.
Further, according to the present invention, the page number information is embedded in the image, and the images scanned and extracted according to the extracted information at the time of reading are collectively stored in one file in the order of the pages, so that the user can handle the file. Convenience is improved.
In addition, according to the present invention, by providing the image similarity determination means, even if there is continuity in the page information between the input image and the image data existing in the storage area, different files are stored in one file. The convenience for the user is improved.

Further, according to the present invention, the page is not rearranged automatically but only with the confirmation of the user, so that a finer operation is possible for the user and the convenience is improved.
In addition, according to the present invention, by performing similarity determination between an input image and an image included in the storage area, which should be different from each other, according to the document type, the clearly input image is stored in the storage area. The accuracy in determining whether it is part of a series of included image files is improved.
In addition, according to the present invention, since the start page of the page information is randomly determined, even if the page information is embedded by different printers, the possibility that they are duplicated is reduced, and the convenience for the user is improved. .
In addition, according to the present invention, the processing speed is improved and the convenience for the user is improved by not performing the similarity determination when there is no page closer to the page number proximity determination means.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
[Example]
As a preferred embodiment of the present invention, a description will be given by taking copying in an electrophotographic MFP (Multi Function Printer: image forming apparatus) as an example.
FIG. 1 is a schematic block diagram of an image forming apparatus according to an embodiment of the present invention.
As shown in FIG. 1, the image forming apparatus includes a scanner unit 1 that reads a document, a storage device 5 that holds image data acquired via the scanner unit 1 and the network 3, and an image captured from the scanner unit 1. And image processing unit 7 for processing image data acquired via network 3 and image data acquired from scanner unit 1 and image data acquired via network 3 either directly or stored in storage device 5 And a printer unit 9 for outputting.

In this image forming apparatus, all the image data obtained by the scanner unit 1 is once held in the storage device 5 and then output by the printer unit 9.
First, an outline of the printer unit 9 will be described. FIG. 2 is a schematic configuration diagram of the printer unit 9 shown in FIG.
Here, an image created by a PC application is converted into bitmap data (hereinafter referred to as image data) by a driver. Further, the image data is subjected to halftone processing so that the printer can output it. The image data after halftone processing is further subjected to the predetermined processing described below.

In FIG. 2, the light source 11 emits light according to cyan image data, and static electricity is generated by the photoelectric effect on the photosensitive member 13, and the cyan developing unit 17 is rotated by rotating the revolver 15 against the photosensitive member 13 charged with static electricity. The toner is charged from the toner to form an image. Similarly, by forming an MYK image, a CMYK image is formed on the photoreceptor 13.
Finally, an electrostatically charged paper transported by the transport belt 19 is brought into close contact with the photosensitive member 13 to form an image on the paper. Since the image created on the paper is merely on the paper, it is fixed by applying a high temperature and high pressure load, and an output image is obtained. Here, since the magnitude of such a load is considerable, when RFID (Radio Frequency Identification) in which page information is described is added, there is a possibility that the RFID loaded on the paper is destroyed.

Next, the operation of the image forming apparatus shown in FIG. 1 will be described with reference to the flowchart of FIG. 3 and FIG. FIG. 3 is an operation flowchart of the image forming apparatus shown in FIG. FIG. 4 is a block diagram showing the configuration of the image processing unit 7 shown in FIG.
First, in step 101 of FIG. 3, when the MFP is operated, a user is specified by an ID card and a password, and an authentication process is performed. Here, it is assumed that the user has a unique user name of 6 ASCII characters. When the user sets an original on the ADF and presses the copy start button, the original supplied by the ADF is optically read by the scanner unit 1, and the image processing unit 7 performs AD conversion to 1 byte per digital pixel. It is converted into a signal (step 103).
Next, the information extraction unit 21 of the image processing unit 7 determines whether or not the information embedded in the image exists, and if it exists, 64-bit information is acquired (step 105). The part 21 will be described in detail later.

Next, as shown in FIG. 5, the user name acquisition unit 23a of the embedding information acquisition unit 23 of the image processing unit 7 reads the upper 48 bits of information corresponding to the user name of the 64-bit image, and follows the ASCII code table. The 48-bit information is converted into 6 ASCII characters (step 107). If there is no 64-bit information, the null character 00 is output for 6 characters. FIG. 5 is a block diagram showing the configuration of the embedded information acquisition unit 23 shown in FIG.
On the other hand, as shown in FIG. 5, the page number acquisition unit 23b of the embedded information acquisition unit 23 reads the lower 16 bits corresponding to the page number out of the 64 bits, converts it into a decimal number, and outputs it. If there is no 64-bit information, 0 is output (step 107).

Next, the embedding information generation unit 25 of the image processing unit 7 generates information to be embedded in the read image (step 109). The information embedded here is the same user name and page number as those acquired by the embedded information acquisition unit 23. The embedded information generation unit 25 will be described later.
The embedded information is stored in the storage device 5 in a pair with the image data obtained by the scanner unit 1 (step 113). Here, the image data is stored together in one page order. At the same time, the page number obtained by the embedding information generation unit 25 is sent to the page order determination unit 27. The page order determination unit 27 determines the page output order based on the page number obtained by the embedding information generation unit 25. Based on this page output order determination, if necessary, rearrangement or group integration is performed (step 111). The page order determination unit 27 will be described later.
In accordance with the page order obtained by the page order determination unit 27, the image data and the embedding information corresponding thereto are read from the storage device 5, and the embedding unit 29 adds the embedding information to the read image data. The image data thus obtained is sent to the printer unit 9 and output from the printer unit 9 together with the image information (steps 113, 115 and 117). The embedding 29 will be described later.
As shown in FIG. 4, the image processing unit 7 includes an information extraction unit 21, an embedded information acquisition unit 23, an embedded information generation unit (page number acquisition unit) 25, and a page order determination unit 27. And an embedded portion 29.
As described above, the point of the present invention is that the rearrangement as described above is achieved by using digital watermarks without embedding RFIDs or barcodes.

As described in the prior art, the RFID may be destroyed by the fixing of the electrophotographic method, and there is a demerit that the cost is high when a large amount of output is assumed. The barcode also has a demerit that it impairs the user's freedom of document editing. On the other hand, since the digital watermark method only processes the document, it can be used with any printing method, and there is no additional cost at the time of output, and there is a concern that the user's freedom of document editing may be impaired. Absent.
Furthermore, another point of the present invention is that not only the information embedded by the digital watermark but also the image similarity determination is used as reference information for page rearrangement. Since the digital watermarking method processes a document, if a large amount of information is embedded, the image quality is greatly deteriorated. If the amount of information to be embedded is reduced in order to maintain good image quality, the amount of information allocated to the page number is naturally reduced. That is, the number of digits that can be added as a page number is reduced. Then, since the size of the number is limited, there is an increased possibility that the document is originally different but is determined to be the same document. Even in such a case, it is possible to prevent the integration of different documents if it is possible to determine the similarity of images and detect that the documents are different.
As described above, by applying the present invention, even when the user sets the page order of a plurality of originals in the ADF, the original page order is output, which greatly reduces the user's trouble. It is possible to reduce.

Next, each component in the image processing unit 7 will be described.
[Embedding information generation unit 25]
Next, the embedded information generation unit 25 will be described in detail. As shown in FIG. 6, the embedded information generating unit 25 includes a user name generating unit 25a, a page number generating unit 25b, and a random number generating unit 25c. FIG. 6 is a block diagram showing the configuration of the embedded information generating unit 25 shown in FIG.
As shown in FIG. 6, when the user name output by the embedded information acquisition unit 23 is not null, the user name generation unit 25a outputs the user name obtained from the embedded information acquisition unit 23 as it is.
On the other hand, when the user name is null, the user name generation unit 25a outputs the user name obtained by the authentication process as the user name. When the page number obtained from the embedding information acquisition unit 23 is not 0, the page number generation unit 25b outputs the page number obtained from the embedding information acquisition unit 23 as it is, while the obtained page number is If it is 0, the page number is generated while incrementing the page number in the input order of the input images from the start to the end of scanning of a plurality of documents. However, the number from the random number generation unit 25c is used for the first page number. Therefore, the start page of the page information is determined at random, and even if the page information is embedded by different printers, the possibility that they will be duplicated is reduced and the convenience for the user is improved.

[Embedding part 29]
Next, the embedding part 29 will be described in detail. As shown in FIG. 7, the embedding unit 29 has a DCT unit 29a, an adding unit 29b, and an inverse DCT unit 29c. FIG. 7 is a configuration block diagram of the embedding unit 29 shown in FIG.
As shown in FIG. 7, the DCT unit 29a performs DCT on every 8 × 8 pixels of the input image as one block to generate a DCT coefficient. In the adding unit 29b, first, the value of the region 2 × 2 pixels indicating the high frequency of the DCT coefficient is set to 0, and 4 bits × 4 blocks are set as one unit, and 1-bit information is added according to the embedding information. By doing so, 64-bit information can be added by (2 pixels × 2 pixels) × (4 blocks × 4 blocks). At this time, the same information is added every time 4 blocks × 4 blocks are repeated. By doing this, even if the image quality deteriorates or is altered by post-output scanning, the same information is embedded in various areas, so that the possibility that the embedded information can be extracted accurately increases. Finally, the inverse DCT unit 29c obtains an output image by performing inverse DCT on the image data to which information is added.

[Embedding information acquisition unit 23]
Next, the embedded information acquisition unit 23 will be described in detail. As shown in FIG. 8, the embedded information acquisition unit 23 includes a DCT unit 23a and a reading unit 23b. FIG. 8 is a configuration block diagram of the embedded information acquisition unit 23 shown in FIG.
As shown in FIG. 8, DCT is performed on the input image data by the DCT unit 23a. In the reading unit 23b, 8 × 8 pixels of the image data after DCT are regarded as one block, and 1-bit information is extracted from each of the 2 × 2 pixels in the high-frequency region. Further, the information read from the 4 × 4 block is integrated to output 64-bit embedded information.

[Page Order Determination Unit 27]
Next, the page order determination unit 27 will be described in detail. As shown in FIG. 9, the page order determination unit 27 includes a storage unit 27a, a page number proximity determination unit 27b, an image similarity determination unit 27c, a confirmation unit 27d, and a page rearrangement unit 27e. Yes. FIG. 9 is a configuration block diagram of the page order determination unit 27 shown in FIG.
As shown in FIG. 9, the input is a page number generated by the embedding information generation unit 25, and the input page number is held in the storage unit 27a. Here, the structure of data held in the storage unit 27a is as shown in the table of FIG. Here, the category indicates the document type determined by the image processing apparatus. The input image assumed in this embodiment is a mixture of a plurality of types of documents, and the category indicates each of the plurality of documents. FIG. 13 is a table showing the structure of data held in the storage unit 27a. The storage unit 27a is assumed to be included in the storage device 5.

Next, the input page number is similarly sent to the page number proximity determination unit 27b. The page number proximity determination unit 27b determines whether each page number already stored in the storage unit 27a is close to the input number. In the present embodiment, the difference is set to 20 in the present embodiment, but other numbers may of course be used and the user may be able to set them. When the page number proximity determining unit 27b determines that there is a similar page number, the image similarity determining unit 27c reads the image of the page from the storage device 5 and determines the image similarity. The image similarity determination unit 27b will be described later. Note that when the page number proximity determining unit 27b determines that there is no similar page number, the image similarity determining unit 27c does not perform image similarity determination.
When the image similarity determination unit 27b determines that the two pages are sufficiently close, the confirmation unit 27d informs the user that the page order is to be changed, prompts the user confirmation, and asks for the user's instruction. In the present embodiment, overlapping of the page order is prevented with high accuracy by combining the digital watermark and the similarity determination of the image. Of course, there is also a possibility that the page order is duplicated. By having the user make a final decision, it is possible to completely prevent page duplication. On the other hand, since the user's operation may become complicated, it is desirable that the user can arbitrarily set the validity of the confirmation unit 27d. When the confirmation unit 27d instructs the user to change the page order, the page rearrangement unit 27e rearranges the pages existing in the storage unit 27a.

  In the page rearrangement unit 27e, when the previous page number proximity determination unit 27b, the image similarity determination unit 27c, or the confirmation unit 27d determines that it does not belong to any existing category, a new storage area is created. Create a category and register the page number there. If it is determined that the category is included in any of the categories, the page numbers are rearranged in ascending order after being inserted into the category. It should be noted that bubble sorting was used as a very general method. Finally, when the scan is completed, the page numbers existing in the storage unit 27a are output in order, such as the page number of category 1 in ascending order and the order of page number in category 2 in ascending order.

そして、特徴量算出部３０ｂでは及び領域分割部３０ａで得られた領域毎に、平均色、エッジ総量、中心位置からのモーメントからなる特徴量を算出し、オブジェクト種判定部３０ｃに送る。オブジェクト種判定部３０ｃは、領域分割部３０ａによって得られた矩形領域がどのようなオブジェクトであるか判定する。ここでいうオブジェクトとは、文字列、背景、その他、である。ここでは後述するサポートベクタマシンを用いて事前に学習しておき、入力画像の特徴からオブジェクト種が推定する。 [Image similarity determination unit 27c]
Next, the image similarity determination unit 27c will be described in detail. As shown in FIG. 10, the image similarity determination unit 27c includes an area division unit 30a, a feature amount calculation unit 30b, an object type determination unit 30c, a form determination unit 30d, a slide determination unit 30e, and a document determination unit. 30f, a comprehensive determination unit 30g, a storage unit 30h, and a similarity determination unit 30i. FIG. 10 is a block diagram showing the configuration of the image similarity determination unit 27c shown in FIG.
As shown in FIG. 10, when two images to be compared, and subsequently the first image and the second image are sequentially input to the image similarity determination unit 27c, first, the first image is the region dividing unit 30a. The area division 30a divides the input image into objects such as a character area and a picture area. As a specific region dividing method, the method described in Non-Patent Document 1 is used.
In outline, first, a white region is extracted as a background by binarization. Here, binarization used simple binarization in which the density is 127 or higher. Next, an image like a chain is obtained by thinning the white area. Further, by removing unnecessary chains that satisfy Equation 1, the boundary of the region can be obtained. In Equation 1, D is a minimum distance, W is an average line width difference, and tD and tW are thresholds set in advance, respectively. The explanation of these terms is left to Non-Patent Document 1, but in this example, tD = 5 and tW = 30 were used. Further, the area dividing unit 30a calculates a rectangle including the boundary.
(Formula 1)

Then, the feature amount calculation unit 30b calculates a feature amount including the average color, the total amount of edges, and the moment from the center position for each region obtained by the region division unit 30a, and sends the feature amount to the object type determination unit 30c. The object type determination unit 30c determines what kind of object the rectangular area obtained by the area division unit 30a is. An object here is a character string, a background, and others. Here, learning is performed in advance using a support vector machine described later, and the object type is estimated from the characteristics of the input image.

Next, each of the form determination unit 30d, the slide determination unit 30e, and the document determination unit 30f uses the composition ratio (three-dimensional) of the object type obtained by the object type determination unit 30c as a feature amount, The document quality is output by a classifier previously learned by a support vector machine. As will be described later, the support vector machine is generally a discriminator that performs a binary determination, but it is possible to calculate how far the input feature amount is from the discriminant plane, and with this, It is a form, a slide, and a document.
The above-described form-likeness, slide-likeness, and document-likeness, that is, the document type, is determined by the overall determination unit 30g, and the most likely one is used as the document type. When the first image is input, the document type is stored by the storage unit 30h, and when the second image is input, the first document type and the first document type read from the storage unit 30h are stored. The similarity determination unit 30i determines whether the document types of the two images match. Note that the storage unit 30 h is included in the storage device 5.

Next, an outline of the support vector machine will be described. The details are described in Non-Patent Document 2, but the outline is as follows. FIG. 11 is an explanatory diagram showing an outline of the operation of the support vector machine (SVM).
First, as a precondition, it is assumed that there are two types of vector groups represented by ○ × as shown on the left side of FIG. SVM can be said to be an algorithm for determining a hyperplane (see the right in the figure) for optimally separating these two classes. In SVM, optimally dividing two vector groups is equivalent to maximizing the ability to respond when an unknown vector is input, that is, the generalization ability. In order to realize this, a vector (Support Vector) existing at the boundary position between two vector groups is found, and the hyperplane is set so that the distance between the Support Vector and the hyperplane is maximized. Here, since there is erroneous teacher data in actual operation, it is necessary to determine a parameter for setting an allowable error amount (soft margin). The above is a description of the linear SVM, but actual teacher data is not necessarily a vector group that can be linearly discriminated. However, by projecting the feature vector onto a higher-order space (kernel trick) and obtaining a hyperplane in that space, it is possible to cope with nonlinear problems.

（数式３）

（数式４）

（数式５）

In order to realize the above, as a result, a Lagrange multiplier vector αi that maximizes Expression 3 under the condition of Expression 2 is obtained using the teacher data (xi, yi). Then, using the teacher data group S corresponding to the non-zero element of the Lagrange multiplier vector elements (this is the support vector) and any one of the teacher data (x0, y0), the hyperplane parameter ω, h is obtained (Equations 4 and 5).
(Formula 2)

(Formula 3)

(Formula 4)

(Formula 5)

以上の様に、埋めこまれたページ番号情報のみでなく、類似画像判定を付加することで、異なる文書のページが重複していた場合でも、正しくページの並べ替えができる可能性が高くなる。更に本実施例で述べた様に、文書タイプを識別し、これにより類似画像判定を行う構成としたが、これは以下の理由による。例えばパタンマッチングなどを用いて画像の類似度判定を行う場合を考えると、得られる類似度は近さを示す量になる。つまり、これが実際に類似しているか否かを判定するためには一種の閾値を用いねば成らない。更に、本実施例では一致する二つの文書の類似度を判定するのではなく、ページの異なる二つの文書の類似度を判定しようとするため、閾値の設定は難しくなる。結果実用的にはユーザが閾値を設定し、試行錯誤の上で決定しなければ成らない。一方、文書タイプ識別は、二値的な判断であり、似ているか似ていないかは文書のタイプが一致するか一致しないかという二値的な判断に成るため、上記の様な問題は発生しない。 In the above formula, K (x, y) represents a kernel function for realizing a kernel trick. Various kernel functions have been devised. In this embodiment, a radial basis function (RBF) is used. RBF is a function expressed by Equation 6, and C is an arbitrary number.
(Formula 6)

As described above, by adding not only the embedded page number information but also similar image determination, there is a high possibility that the pages can be correctly rearranged even when pages of different documents are duplicated. Further, as described in the present embodiment, the document type is identified and the similar image determination is performed based on the document type. This is for the following reason. For example, considering the case where image similarity determination is performed using pattern matching or the like, the obtained similarity is an amount indicating closeness. In other words, in order to determine whether or not they are actually similar, a kind of threshold value must be used. Furthermore, in this embodiment, since the similarity between two matching documents is not determined, but the similarity between two documents with different pages is determined, it is difficult to set a threshold value. As a result, in practice, the user must set a threshold value and determine it through trial and error. On the other hand, the document type identification is a binary decision, and the above-mentioned problems occur because it is a binary decision whether the document type matches or does not match. do not do.

Next, a case where scanning is performed using the embodiment of the image forming apparatus illustrated in FIG. 1 will be described. FIG. 12 is a flowchart of a scanning operation using the embodiment of the image forming apparatus shown in FIG.
It should be noted that the scanned image data stored here is not necessarily managed for each page, but a plurality of pages such as multi-page Tiff or PDF are managed in one file. To do.
In step 201 of FIG. 12, after the original is optically read by the scanner unit 1, it is AD-converted by the image processing unit 7 and converted into RGB image data of 8 bits per pixel and embedded in the image data. 64-bit embedded information is extracted (step 203). Next, information representing the user name and information representing the page number are extracted from the extracted information (step 205), and the input document has already been stored in the storage device 5 based on the extracted page number. It is determined whether it is a part of the image data (step 207).
When it is determined that the document input in the page order determination is a part of the image data already stored in the storage device 5, the document image appropriately input to the image data already existing in the storage device 5 Are integrated and stored in the storage device 5 (steps 209 and 211).

  With the above configuration, the user is freed from various operations during scanning. For example, as described above, even if a plurality of mixed documents are not arranged and rearranged, the files held in the storage means are arranged for each document and in the correct page order. Furthermore, even if only one page of the document set in the ADF is missing, the page is automatically inserted into the correct document in the correct page order by scanning that page later.

1 is a schematic block diagram of an image forming apparatus according to an embodiment of the present invention. It is a schematic block diagram of the printer part 9 shown in FIG. 3 is an operation flowchart of the image forming apparatus illustrated in FIG. 1. FIG. 2 is a block diagram illustrating a configuration of an image processing unit 7 illustrated in FIG. 1. FIG. 5 is a configuration block diagram of an embedded information acquisition unit 23 illustrated in FIG. 4. FIG. 5 is a configuration block diagram of an embedded information generation unit 25 shown in FIG. 4. FIG. 5 is a configuration block diagram of an embedding unit 29 shown in FIG. 4. FIG. 5 is a configuration block diagram of an embedded information acquisition unit 23 illustrated in FIG. 4. FIG. 5 is a configuration block diagram of a page order determination unit 27 shown in FIG. 4. FIG. 10 is a configuration block diagram of an image similarity determination unit 27c shown in FIG. 9. It is explanatory drawing which showed the outline | summary of operation | movement of a support vector machine (SVM). 3 is a flowchart of a scanning operation using the embodiment of the image forming apparatus illustrated in FIG. 1. It is a figure which shows the structure of the data hold | maintained at the memory | storage part 27a with a table | surface.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Scanner part, 3 ... Network, 5 ... Memory | storage device, 7 ... Image processing part, 9 ... Printer part, 11 ... Light source, 13 ... Photoconductor, 15 ... Revolver, 17 ... Developing unit, 19 ... Conveyor belt, 21 ... Information extraction unit, 23 ... Embedded information acquisition unit, 23a ... User name acquisition unit, 23b ... Page number acquisition unit, 25 ... Embedded information generation unit, 25a ... User name generation unit, 25b ... Page number generation unit, 25c ... Random number generation unit, 27... Page order determination unit, 27a... Storage unit, 27b... Page number proximity determination unit, 27c... Image similarity determination unit, 27d ... confirmation unit, 29 ... embedding unit, 29a ... DCT unit, 29b. Addition unit, 29c ... inverse DCT unit, 30a ... area division unit, 30b ... feature amount calculation unit, 30c ... object type determination unit, 30d ... form determination unit, 30e ... slide determination unit, 30f ... document determination unit, 30g Comprehensive determination unit, 30h ... storage unit, 30i ... similarity determination unit

Claims (7)

An image forming apparatus that reads a plurality of pages of an original to form an image,
Image reading means for reading the original and generating image data;
An information embedding means for embedding information having at least the page number of the original in a digital watermark method in the original read by the image reading means;
Information extracting means for extracting information embedded in the manuscript;
Page number acquisition means for acquiring the page number of the document from the information read by the information extraction means;
Storage means for storing a plurality of pages of image data read by the image reading means as one group;
A page number proximity determining unit that determines whether the page number acquired by the page number acquiring unit and each page number already stored in the storage unit are close to each other, and a page closer to the page number proximity determining unit An image similarity determination unit that determines the image similarity of the image data of the plurality of pages when it is determined to exist, and the pages of the document are rearranged according to the image similarity determined by the image similarity determination unit. An image forming apparatus comprising: a page ordering unit having a page rearranging unit for grouping the pages of the document . The image forming apparatus according to claim 1 , wherein the page order determination unit includes a confirmation unit that confirms a user when rearranging the pages of the document or integrating the group of document pages. apparatus. The image similarity determining means image forming apparatus according to claim 1 or 2, characterized in that it has a document type identifying means for determining a document type of page of at least the document. An embedding information generating means for randomly determining the first page number of image data composed of a plurality of pages read by the image reading means when the page number acquired by the page number acquiring means is 0 ; The image forming apparatus according to claim 1 , wherein the image forming apparatus is an image forming apparatus. An image processing method in an image forming apparatus for reading a plurality of pages of an original to form an image,
Reading the document by an image reading unit to generate image data;
Embedding information having at least the page number of the original in an electronic watermark method in the original read by the image reading means by the information embedding means;
Extracting information embedded in the original by information extraction means;
Acquiring a page number of the document from information read by the information extraction unit by a page number acquisition unit;
Storing a plurality of pages of image data read by the image reading means by a storage means as one group;
Determining whether the page number acquired by the page number acquisition means by the page number proximity determination means and each page number already stored in the storage means are close numbers;
A step of determining image similarity of the image data of the plurality of pages by an image similarity determination unit when the page number proximity determination unit determines that there is a close page;
And a step of rearranging the pages of the manuscript according to the degree of similarity of the image determined by the image similarity determining unit, and integrating the pages of the manuscript into a group. Processing method. A program for causing a computer to execute the image processing method according to claim 5 . An information recording medium on which a computer program for achieving the image processing method according to claim 5 is recorded.

Images