JP4192887B2 - Tamper detection device, watermarked image output device, watermarked image input device, watermarked image output method, and watermarked image input method - Google Patents

Description

  The present invention relates to detection of falsification of an image such as a document image.

  When print data is printed on a print medium such as copy paper and a printed matter is generated, the contents printed on the printed matter can be grasped by looking at texts and images printed on the printed matter.

  On the other hand, there exists “digital watermark” that embeds information for preventing copy / counterfeiting or confidential information in digital data such as images and document data in an invisible format. There is a technique for preventing illegal acts such as forgery by embedding the “digital watermark” in document data or the like.

  When printing the print data on the print medium by applying the above-mentioned “digital watermark” technology, if the digital watermark corresponding to the print data is printed together with the print data, It is possible to discover falsification of the printed matter such as alteration of characters printed on the printed matter from the electronic watermark information printed on the printed matter, just by visual observation (see, for example, Patent Document 1). Note that whether or not tampering has occurred is determined by comparing the print result with that printed with a digital watermark.

JP 2000-232573 A

  However, in order to determine whether or not the printed material has been tampered with, it is necessary to visually compare the printed content extracted from the digital watermark with the printed content printed on the printed material.

  Therefore, there are problems as shown in (1) and (2) below. (1) Since it is a visual determination, it has been difficult to process whether a large amount of printed material has been tampered with in a short time. (2) Since it is necessary to read and compare the printed contents one character at a time, there is a possibility that alterations may be missed due to human error.

  The present invention has been made in view of the above problems, and an object of the present invention is to automatically embed watermark information, generate an output image, extract the embedded watermark information from the output image, and output it. Provided are a new and improved tamper detection device, watermarked image output device, watermarked image input device, watermarked image output method, and watermarked image input method capable of automatically detecting image tampering. That is.

  In order to solve the above-described problem, according to a first aspect of the present invention, a tamper detection device is provided. The tampering detection apparatus includes: an image processing unit that inputs an image and executes a process of cutting out a specific region of a region for determining whether one or more tampering has been performed from the input image; and the image processing unit A watermark information embedding unit that embeds at least one of region information relating to the position and / or size of the specified region or a partial image of an image corresponding to the specific region as watermark information and generates an output image; A watermark information extraction unit that extracts watermark information embedded in the input output image; an image transformation unit that processes the input output image and transforms the input image into an input image having substantially the same size as the image; A pseudo image generation unit that generates a pseudo image that is at least similar to a partial image of a specific region existing in the image; and a differential image that includes a difference between the input image and the pseudo image Generated, if the difference region in said difference image is present, is characterized by comprising a determining alteration determination unit and an output image to be deformed original input image has been tampered with.

  According to the present invention, only a partial image corresponding to a specific area is cut out from the image, and the partial image, area information regarding the specific area, and the like are embedded in the image as watermark information corresponding to, for example, an electronic signature. An image in which watermark information is embedded is generated as an output image. When judging whether or not the output image has been tampered with, the input image is input and input by executing processing such as correction, binarization of the output image, or enlargement / reduction of the output image. If an image is generated, a pseudo image is generated based on the watermark information, and a difference area exists in the difference image obtained by taking the difference between the two images, it can be determined that the image has been tampered with. According to such a configuration, the apparatus can automatically detect tampering detection without depending on human eyes, and a tamper detection process can be executed quickly and accurately for a large number of output images. Also, the watermark information embedded in or extracted from the image embeds the image or area information related to only the area (specific area) subject to the falsification detection process, so the data amount is reduced compared to the image data of the entire area of the image. This makes it possible to improve the efficiency and speed of the falsification detection process. The difference area is an area that appears when, for example, the image of the input image and the pseudo image corresponding to each other are different from each other.

  In order to solve the above problems, according to another aspect of the present invention, a watermarked image output apparatus is provided. The watermarked image output apparatus includes: an image processing unit that inputs an image and executes a process of cutting out a specific region of the region for determining whether or not one or more of the input image has been tampered with; A watermark information embedding unit that embeds at least one of region information relating to the position and / or size of the specific region cut out by the above, or a partial image of an image corresponding to the specific region as watermark information, and generates an output image. It is characterized by providing.

  According to the present invention, only a specific region is cut out from an image, and a partial image of the image related to the specific region, region information of the specific region, and the like are embedded in the image as watermark information corresponding to, for example, an electronic signature. The embedded watermark information plays a role of electronic signature in the falsification detection process and automatically detects falsification by comparing the pseudo image generated from the watermark information with the input image generated from the output image. Processing is executed. With this configuration, the apparatus can automatically detect falsification by embedding watermark information in an image. Further, by narrowing down the target of the falsification detection process in the image, it is possible to reduce the data amount of the watermark information embedded in the image.

  The image processing unit executes a process of cutting out the specific area, and further executes a reduction process of reducing a partial image of the image corresponding to the cut out specific area at a predetermined reduction rate; Information, a partial image of the image corresponding to the specific area, a reduced image obtained by reducing the partial image of the image corresponding to the specific area, or a reduction ratio of the reduced image is embedded as watermark information. Also good. With this configuration, it is possible to further compress and reduce the amount of data compared to simply embedding watermark information relating to a specific area in an image.

  The image processing unit executes a process of cutting out the specific area, and further reduces the partial image of the image corresponding to the cut out specific area at a reduction rate determined according to the importance for each specific area. May be configured to execute. With this configuration, it is possible to improve the detection accuracy for particularly important areas among the specific areas to be detected for alteration, while improving the processing efficiency by increasing the reduction ratio for the specific areas where the priority is lowered. be able to.

  The image processing unit may be configured to dynamically change the importance set for each specific area. According to such a configuration, the tampering process can be efficiently performed without determining the importance in advance by dynamically changing the importance according to the symbols or the like existing in the specific area.

  The specific area may be configured to be a character area including at least one of a character, a symbol, or a figure. In addition, the said character is 1 or 2 characters or more, and a symbol and a figure are 1 or 2 or more.

  The image processing unit may be configured to batch one or more reduced images and generate one composite image. With this configuration, watermark information embedded in an image can be further compressed, and the amount of data can be reduced.

  The watermarked image output apparatus further includes a straight line / dashed line processing unit for extracting a straight line and / or broken line region having a predetermined length included in the image, and the watermark information embedding unit is extracted by the straight line / dashed line processing unit. The straight line area information regarding the position and / or size of the straight line and / or broken line area can be further embedded in the image as watermark information.

  In order to solve the above problems, according to another aspect of the present invention, a watermarked image input apparatus is provided. The watermarked image input device includes at least one of region information regarding the position and / or size of one or more specific regions for determining whether or not the image has been tampered with, or a partial image of an image related to the specific region. A watermark information extraction unit that inputs an output image generated by embedding one in the image as watermark information and extracts the watermark information embedded in the input output image; and processes the input output image; An image deforming unit that transforms an input image having substantially the same size as the image; a pseudo image generating unit that generates a pseudo image at least similar to an image in a specific region existing in the image region of the original image; and an input image and a pseudo image Tampering determination that generates a difference image consisting of differences and determines that the output image that is the source of deformation of the input image has been tampered with when a difference area exists in the difference image It is characterized in that it comprises and.

  According to the present invention, by inputting an output image, watermark information embedded in the image is extracted, and processing such as correction of the output image, binarization of the output image, or enlargement / reduction of the output image is performed. To generate an input image, generate a pseudo image based on the watermark information, and if there is a difference area in the difference image obtained by taking the difference between the two images, determine that the image has been tampered with Can do. According to such a configuration, the apparatus can automatically detect tampering detection without depending on human eyes, and a tamper detection process can be executed quickly and accurately for a large number of output images. In addition, since the watermark information extracted from the image embeds the image or area information related to only the area (specific area) subject to the falsification detection process, the amount of data can be reduced compared to the whole area of the image. Can be made more efficient and faster. The difference area is an area that appears when, for example, the input image and the pseudo image have a one-to-one corresponding image area.

  The watermark information extraction unit includes at least one of area information, a partial image of an image relating to the specific area, a reduced image obtained by reducing a partial image of the image relating to the specific area, or a reduction ratio of the reduced image as watermark information. An output image generated by being embedded in the input image may be input, and the watermark information embedded in the input output image may be extracted.

  When the pseudo image generation unit generates a background image having substantially the same size as the image, and the watermark information includes a partial image of the image corresponding to the specific area, the image is directly arranged in the background image according to the area information, or the watermark information When a reduced image corresponding to a specific area is included in the image, the reduced image is first enlarged at a reciprocal of the reduced reduction rate, the enlarged reduced image is arranged according to the area information, and a pseudo image is generated. The image may be configured to be similar to the image.

  The pseudo image generation unit is simulated based on at least one of area information on the position and / or size of the specific area included in the extracted watermark information, an image on the specific area, or a reduced image obtained by reducing the image on the specific area. By generating an image, the pseudo image may be configured to be similar to the image.

  The specific area may be configured to be a character area including at least one of a character, a symbol, or a figure. In addition, the said character is 1 or 2 characters or more, and a symbol and a figure are 1 or 2 or more.

  The image transformation unit cuts out an area that matches the specific area in the input output image based on the area information regarding the position and / or size of the specific area included in the watermark information, and converts the area into the watermark information corresponding to the specific area. The image of the clipped area is reduced at the included reduction ratio, and the image of the clipped area is enlarged at a magnification that is the reciprocal of the reduction ratio, and is then placed at the original position in the output image. May be. With such a configuration, it is possible to improve the accuracy of tamper detection by setting the same conditions as the generation conditions of the generated pseudo image and making the image quality of the pseudo image and the image quality of the input image substantially the same, and whether or not the tampering has been performed more accurately. Can be determined.

  The image transformation unit reduces the image of the clipped region at a reduction rate determined according to the importance for each specific region, and further enlarges the image of the clipped region at a magnification that is the inverse of the reduction rate. , It may be arranged at the original position in the output image.

  The watermark information extraction unit included in the watermarked image input unit extracts, as watermark information, straight line / broken line area information relating to the position and / or size of a straight line and / or broken line area having a predetermined length included in the image from the output image. The watermarked image input apparatus further includes a straight line / broken line removal unit for removing straight lines and / or broken line areas present in the input output image based on straight line / broken line area information included in the watermark information. You may comprise.

  In order to solve the above-described problem, according to another aspect of the present invention, an image is input, and a process of cutting out a specific area of an area for determining whether one or more tampering has been performed from the input image. Image processing to be executed; and embedding at least one of region information regarding the position and / or size of the specific region cut out by the image processing processing, or a partial image of the image corresponding to the specific region as watermark information, And a watermark information embedding process for generating an output image.

  In order to solve the above-described problem, according to another aspect of the present invention, region information regarding the position and / or size of one or more specific regions for determining whether or not an image has been tampered with, or a specific region A watermark information extraction process for inputting an output image generated by embedding at least one of the partial images of the image as watermark information in the image and extracting the watermark information embedded in the input output image; An image transformation process that processes the input output image and transforms the input image into an input image having substantially the same size as the image; a pseudo image generation process that generates a pseudo image that is at least similar to a partial image of a specific area existing in the image; If a difference image is generated that consists of the difference between the input image and the pseudo image, and the difference area exists in the difference image, the output image that is the source of deformation of the input image has been altered. It is characterized by comprising a determining alteration determination process.

  As described above, according to the present invention, the falsification detection device can automatically detect whether or not falsification exists in a document image without detecting the presence or absence of falsification by human eyes. it can. In addition, since tampering is detected by an apparatus instead of human eyes, the accuracy of detection processing for determining whether or not tampering exists in a document image can be improved. On the other hand, watermark information is embedded in the document image in order to detect falsification, but the watermark information is limited to only the area of the document image that is to be examined for prevention of falsification, not the entire area of the document image. Since the information about the area is embedded, the amount of data can be reduced.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described in detail with reference to the accompanying drawings. In the following description and the accompanying drawings, components having substantially the same functions and configurations are denoted by the same reference numerals, and redundant description is omitted.

  The document image described below is described on the assumption that the document image is black and white (background is white, character is black). However, the present invention is not limited to this, and the document image background and character color are different. , Any background color combination can be implemented.

(About the first embodiment)
First, the document output unit 10 provided in the falsification detection apparatus according to the first embodiment will be described with reference to FIG. FIG. 1 is a block diagram illustrating a schematic configuration of a document output unit 10 according to the first embodiment.

  As shown in FIG. 1, the document output unit 10 receives a document image (image) 11 and outputs an output document image 14 with a watermark. The document image 11 is image data obtained by imaging document data created by document creation software such as word processing software.

  The output document image 14 is image data in which the features of the document image 11 are embedded as watermark information. Normally, the output document image 14 is printed on a print medium such as paper and stored or distributed. Data may be stored or distributed as it is. The output document image 14 is output by an output device (not shown) corresponding to an output device such as a printer.

  The document output unit 10 includes a document feature data conversion unit (image processing unit) 12 that extracts a feature amount serving as an index indicating the feature of the document image 11 from the document image 11 and converts it into data, and the converted feature amount. A watermark information synthesis unit (watermark information embedding unit) 13 embedded in the document image 11 as watermark information is provided.

  The document feature data conversion unit 12 extracts only the image (partial image) corresponding to the character area existing in the document image 11, reduces the image for each partial image corresponding to the character area, and collects the reduced character areas. By using the image (partial image) as the feature data of the document image, the data amount can be reduced as compared with the case where the entire document image 11 is reduced as it is to be feature data even at the same reduction ratio. In the following description, the character area existing in the document image 11 refers to, for example, the area itself indicating the size or range, or the document image (partial image) corresponding to the area of the character area. It may be pointed out.

  The watermark information combining unit 13 creates a watermarked document image by superimposing the document image 11 and the watermark information. The watermark information synthesizer 13 digitizes the watermark information and converts it into a numerical value, and performs N-element coding (N is 2 or more), and assigns each encoded symbol to a signal prepared in advance. A signal represents a wave having an arbitrary direction and wavelength by arranging dots in a rectangular area of an arbitrary size, and a symbol is assigned to the direction and wavelength of the wave. A watermarked document image is one in which these signals are arranged on the image according to a certain rule.

  Next, the document input unit 20 provided in the falsification detection apparatus according to the first embodiment will be described with reference to FIG. FIG. 2 is a block diagram illustrating an example of a schematic configuration of the document input unit according to the first embodiment.

  The document input unit 20 determines the presence or absence of falsification from the input document image 21 generated by directly or indirectly capturing the output document image 14 embedded with a watermark, and outputs the result as a falsification detection result 26. .

  When the output document image 14 is a printed material, the input document image 21 is taken into a computer (tamper detection device) by an input device (not shown) such as a scanner. Alternatively, when the output document image 14 is still digital data, it is directly captured from the input device by wire or wireless. That is, the input document image 21 and the output document image 14 are the same.

  As shown in FIG. 2, the document input unit 20 includes a watermark information extraction unit 22 that extracts watermark information from the input document image 21, and a document image 11 in which the watermark information extracted from the input document image 21 is embedded by the document output unit 10. A pseudo document image generation unit 23 that restores the feature information indicating the feature, and an input image transformation unit (image transformation unit) 24 that transforms the input document image 21 so as to be compared with the feature of the document image 11 based on the restored feature information. And a tampering detection unit (tampering determination unit) 25 that compares the images calculated by the document image restoration unit 23 and the input image transformation unit 24 to determine tampering. The reason why the input image deformation unit 24 deforms the input document image 21 is to determine whether or not the input document image 21 has been tampered with.

  Next, a series of operations of processing of the document output unit 10 according to the first embodiment will be described. In the following description, the case where the document is horizontally written in the document image 11 will be described. However, the present invention is not limited to such a case, and can be implemented even when the document is vertically written. When the document is written vertically, the document image 11 is rotated 90 degrees to the left or right, and then the same processing as that when the document is written horizontally is executed.

  First, the processing of the document feature data converting unit 12 will be described with reference to FIG. FIG. 3 is a flowchart showing an example of an outline of processing of the document feature data converting unit according to the first embodiment.

  As shown in FIG. 3, the document feature data conversion unit 12 first executes a character area extraction process (S31) for extracting only a character area from the document image 11.

(About document image 11)
Here, the document image 11 according to the present embodiment will be described with reference to FIG. FIG. 4 is an explanatory diagram showing an example of a schematic configuration of a document image according to the present embodiment.

  The document image 11 is data including font information and layout information, and is created by document creation software or the like. The document image 11 can be created for each page as an image in a state where the document is printed on paper. The document image 11 is a black and white binary image. On the image, white pixels (pixels having a value of 1) are backgrounds, and black pixels (pixels having a value of 0) are character regions (regions to which ink is applied). Suppose that

  As shown in FIG. 4, the document image 11 can exemplify an important document such as a certificate that needs to maintain confidentiality, integrity, and the like. Each location where the document is printed on the document image 11 corresponds to a character area. In this embodiment, the character region is a black pixel region because the character is black, but is not limited to this example.

(Character area extraction processing (S31))
The character area extraction process (S31) will be described more specifically with reference to FIGS. 5 to 8 are explanatory diagrams showing an outline of the character region extraction processing.

  In the character region extraction process (S31), character regions are extracted according to the following Step 1 to Step 3.

(Step 1)
First, an expansion process is performed on an image (partial image) corresponding to a character area (black pixel area) in the area of the document image 11. The expansion process uses a general image processing method. FIG. 5 shows an example of the result of performing the expansion process on the character area of the document image 11 shown in FIG.

(Step 2)
When the document feature data conversion unit 12 performs the expansion process on the partial image corresponding to the character area of the document image 11 illustrated in FIG. 4, the document feature data conversion unit 12 labels the partial image corresponding to the character area on which the expansion process has been performed. , Extract the minimum rectangle that encloses each character area.

  The labeling is a so-called black pixel grouping process in which a set of continuously collected black pixels is defined as one group. By labeling, a character area consisting of a minimum rectangle consisting of one or two characters can be extracted.

  When the document feature data conversion unit 12 performs an expansion process on the document area 51 and performs a labeling process, as shown in FIG. 6, the document feature data conversion unit 12 has as few characters as possible, such as one or two characters. Recognizes the character area consisting of the smallest rectangle. FIG. 6 is an explanatory diagram showing an example of an outline of labeling processing for the character region 51 of the document image 11 shown in FIG.

  The document feature data conversion unit 12 performs the expansion process in order to improve the efficiency of the labeling process, and the processing time of the labeling process can be shortened by the expansion process. Note that in the case of a falsification detection apparatus having a sufficiently high calculation speed, “Step 1” may be omitted, and the document image 11 may be directly labeled.

(Step 3)
Next, the document feature data conversion unit 12 grasps the position of the character region consisting of the minimum rectangle that has been labeled, and combines the partial images of the character regions consisting of the minimum rectangle existing on the left and right of the character region. .

As shown in FIG. 6, for example, in the character area 51-1, the coordinate value of the upper left vertex of the rectangle A (third from the rightmost, middle rectangle) is (X _a0 , Y _a0 ), and the coordinate of the lower right vertex. It is assumed that the value is (X _a1 , Y _a1 ). When the coordinate value of the lower right vertex of the rectangle B existing on the left side of this rectangle is (X _b1 , Y _b1 ), the rectangle A and the rectangle B are combined if “X _a0 −X _b1 <Tw” is satisfied. To do. Further, when the coordinate value of the upper left vertex of the rectangle C existing right next to the rectangle A is (X _c0 , Y _c0 ) and satisfies “X _c0 -X _a1 <Tw”, the rectangle A and the rectangle C are combined. .

  As a result, the document feature data converting unit 12 can extract a character area 51 including a plurality of characters, such as the character area 51-1 and the character area 51-2 shown in FIG. As a result of the above combining process, the character area 51 may contain only one character. By combining the minimum rectangles, for example, the subsequent processes such as reduction of the partial image corresponding to the character area can be efficiently performed to some extent without repeatedly executing subsequent processes. FIG. 7 is an explanatory diagram schematically showing the result of the joining process for joining the minimum rectangles shown in FIG.

  Further, the document feature data conversion unit 12 executes the character region extraction process (S31) including Step 1 to Step 3 not only for the character region 51 shown in FIG. 6 but also for other character regions included in the document image 11.

  As shown in FIG. 8, when the document feature data converting unit 12 executes the character area extraction process (S31), the document image 11 includes a plurality of character areas such as areas 1 to 9,. I understand. Note that the document feature data conversion unit 12 manages the position information of each character area using, for example, coordinates. FIG. 8 is an explanatory diagram showing a schematic configuration of the document image 11 as a result of executing the character region extraction (S31).

(Regarding character area cutout / reduction processing (S32))
Next, the document feature data conversion unit 12 cuts out the character area extracted in the character area extraction process (S31) from the document image 11, and reduces each of the cut out character areas (S32).

  The process (S32) for cutting out and reducing the character area will be described with reference to FIG. Note that FIG. 9 is an explanatory diagram illustrating an example of an outline of the cutout / reduction processing (S32) for the character areas of the area 1 and the area 2 illustrated in FIG.

  As shown in FIG. 9, the document feature data conversion unit 12 first cuts out the character areas of the area 1 and the area 2 from the document image 11 shown in FIG. Note that the image data composed of the character areas of the cut out areas 1 and 2 is on the left side shown in FIG.

  Next, the document feature data conversion unit 12 reduces each document image (image data) composed of the character areas of the area 1 and the area 2 (the document image on the right side shown in FIG. 9). Therefore, since the image data is reduced, the data amount of the image data in the area 1 and the area 2 is reduced. The reduction rate will be described by taking a case where the reduction rate is fixed at a predetermined rate as an example. However, the reduction rate is not limited to this example, and the reduction rate may be dynamically changed depending on the character area. Details will be described later.

  The document feature data conversion unit 12 executes a reduction process for all the character areas extracted in the character area extraction process (S31) other than the areas 1 and 2 (S32). The reduced character area is specified as “reduced character area”.

  Further, when extracting the character area, the document feature data converting unit 12 assigns an identification number according to the coordinate value of the upper left vertex of each character area. The assignment method is common between the document output unit 10 and the document input unit 20. For example, in the identification numbers shown in FIG. 8, “1”, “2”,... Are identification numbers such as region 1, region 2,. After setting (0, 0), identification numbers are assigned in ascending order with the first y-coordinate value of the top left vertex of the character area being the first priority and the second x-coordinate value being the second priority. However, it is not limited to such an example.

(About reduced character area composition (S33))
Next, when the character area segmentation / reduction process (S32) is completed, the document feature data conversion unit 12 synthesizes the reduced character areas so as to combine the reduced character areas, thereby generating one image (synthesized image). To do.

  Here, the composite image according to the present embodiment will be described with reference to FIGS. 10 and 11. 10 and 11 are explanatory diagrams showing an example of a schematic configuration of a reduced character region composite image according to the present embodiment.

  The document feature data conversion unit 12 arranges a plurality of reduced character areas so that the reduced character areas are seamlessly connected in a predetermined order. When the reduced character area is single, the above composition process (S33) is not executed and is omitted.

  When the arrangement of all the reduced character areas by the document feature data converting unit 12 is completed, the document feature data converting unit 12 generates a composite image using the arranged reduced character areas as one image data.

  The composite image shown in FIG. 10 includes reduced character area 1 to reduced character area n obtained by reducing areas 1 to n (for example, n is a positive integer) shown in FIG. , 3,...) Are image data when they are combined by seamlessly arranging them vertically (vertical direction).

  In the case of the composite image shown in FIG. 10, the width of the composite image depends on the reduced character region having the largest width among all the reduced character regions 1 to n. The height of the composite image is the total sum of the heights of all the reduced character areas.

  Further, the present invention is not limited to the composite image shown in FIG. 10, and for example, the composite image shown in FIG. 11 can be exemplified as a modification. The composite image shown in FIG. 11 is obtained by fixing the width of the composite image to a predetermined value in ascending order from the reduced character region 1 to the reduced character region 1 to the reduced character region n obtained by reducing the regions 1 to n shown in FIG. This is image data in a case where the reduced character areas are combined by seamless arrangement. That is, if there is a remainder in the width of the composite image, the margin area can be reduced by placing the next reduced character area in the remainder.

As the synthesis processing of the composite image shown in FIG. 11, previously determined width W _s of the composite image from such total area of the reduced character region in advance, as long as it does not exceed the W _s is arranged reduced character area next to the W _s In the case of exceeding, a process of combining by a method such as arranging a reduced character area at the left end of the next stage of the combined image can be exemplified.

  The type of composition processing pattern to be selected may be determined in advance, or a pattern for which the size of the composite image or the size when the composite image is compressed is calculated and selected for each document image 11. May be. In this case, the combined pattern is recorded in a field other than the image feature data.

  The composite image may be embedded in the document image 11 as it is, or may be embedded in the document image 11 after being compressed by a known image compression method.

(Regarding the processing of the watermark information synthesis unit 13)
The watermark information synthesis unit 13 embeds data for falsification detection (falsification detection data Data) in the document image 11 as watermark information. It is assumed that the falsification detection data Data is the following data.

Data 0: Header information such as the size and data size of the document image 11 Data 1: Image data of the composite image or image data obtained by compressing the composite image Data 2: Position information of the character area (extracted by the character area extraction process (S31))
Data3: Composite image creation method ID (when the composite image creation method is changed for each document image 11)

  Note that the falsification detection data Data0, falsification detection data Data1, and falsification detection data Data2 are essential for falsification detection, but the falsification detection data Data3 is appropriately embedded as necessary. The table shown in FIG. 12 is an example of a table related to the falsification detection data Data2, and the coordinate value indicating the position where the character area exists in the document image 11 and the area number have a one-to-one correspondence. In the alteration detection data Data2, the x coordinate and y coordinate of the upper left vertex of the character area rectangle, and the x coordinate and y coordinate of the lower right vertex are recorded. FIG. 12 is an explanatory diagram showing an example of a schematic configuration of a table of falsification detection data Data2 embedded in the document image 11.

  Here, a process of embedding watermark information of the watermark information synthesizing unit 13 will be described with reference to FIG. FIG. 13 is a flowchart showing an example of a schematic flow of processing of the watermark information synthesis unit 13.

  As shown in FIG. 13, first, falsification detection data Data is converted into an N-element code (step S101). N is arbitrary, but in the present embodiment, N = 2 for ease of explanation. Accordingly, the code generated in step S101 is a binary code, and is expressed by a bit string of 0 and 1. In this step S101, the data may be encoded as it is, or the encrypted data may be encoded.

  Next, a watermark signal is assigned to each symbol of the code word (step S102). The watermark signal represents a wave having an arbitrary wavelength and direction by an arrangement of dots (black pixels). The watermark signal will be further described later.

  Further, a signal unit corresponding to the bit string of the encoded data is arranged on the document image 11 (step S103).

  The watermark signal assigned to each symbol of the code word in step S102 will be described. FIG. 14 is an explanatory diagram showing an example of a watermark signal.

  Let the width and height of the watermark signal be Sw and Sh, respectively. Sw and Sh may be different, but in the present embodiment, Sw = Sh is set for ease of explanation. The unit of length is the number of pixels. In the example of FIG. 14, Sw = Sh = 12. The size when these signals are printed on the paper surface depends on the resolution of the watermark image. For example, the watermark image is an image of 600 dpi (dot per inch: unit of resolution, number of dots per inch). 14, the width and height of the watermark signal in FIG. 14 are 12/600 = 0.02 (inch) on the printed document.

  Hereinafter, a rectangle having a width and a height of Sw and Sh is referred to as a “signal unit” as one signal unit. In FIG. 14A, the distance between dots is dense in the direction of arctan (3) (arctan is an inverse function of tan) with respect to the horizontal axis, and the wave propagation direction is arctan (−1/3). . Hereinafter, this signal unit is referred to as unit A. In FIG. 14B, the distance between dots is dense in the direction of arctan (−3) with respect to the horizontal axis, and the wave propagation direction is arctan (1/3). Hereinafter, this signal unit is referred to as unit B.

  FIG. 15 is a cross-sectional view of the change in the pixel value in FIG. 14A as viewed from the direction of arctan (1/3). In FIG. 15, the portion where the dots are arranged is the antinode of the minimum value of the wave (the point where the amplitude is maximum), and the portion where the dots are not arranged is the antinode of the maximum value of the wave.

  In addition, since there are two regions in each unit where dots are densely arranged, the frequency per unit is 2 in this example. Since the wave propagation direction is perpendicular to the direction in which the dots are densely arranged, the wave of unit A is arctan (-1/3) with respect to the horizontal direction, and the wave of unit B is arctan (1/3). Become. Note that a × b = −1 when the direction of arctan (a) and the direction of actan (b) are perpendicular.

  In the present embodiment, symbol 0 is assigned to the watermark signal expressed by unit A, and symbol 1 is assigned to the watermark signal expressed by unit B. These are called symbol units.

  In addition to the watermark signals shown in FIGS. 14 (1) and (2), for example, dot arrangements as shown in FIGS. 16 (3) to (5) are conceivable. In FIG. 16 (3), the distance between dots is dense in the direction of arctan (1/3) with respect to the horizontal axis, and the wave propagation direction is arctan (-3). Hereinafter, this signal unit is referred to as unit C.

  In FIG. 16 (4), the distance between dots is dense in the direction of arctan (−1/3) with respect to the horizontal axis, and the wave propagation direction is arctan (3). Hereinafter, this signal unit is referred to as unit D. In FIG. 16 (5), the distance between dots is dense in the direction of arctan (1) with respect to the horizontal axis, and the wave propagation direction is arctan (−1). In FIG. 16 (5), it can be considered that the distance between dots is dense in the direction of arctan (−1) with respect to the horizontal axis, and the wave propagation direction is arctan (1). Hereinafter, this signal unit is referred to as unit E.

  In this way, in addition to the combinations assigned previously, there can be a plurality of patterns of combinations of units to which symbols 0 and 1 are assigned. Therefore, it is a third party to keep secret which watermark signal is assigned to which symbol. It is also possible to make it impossible to easily decipher the embedded signal.

  Further, when the falsification detection data Data is encoded with a quaternary code in step S102 shown in FIG. 13, for example, the code A symbol 0 is assigned to the unit A, the symbol 1 is assigned to the unit B, and the symbol is assigned to the unit C. 2 and symbol 3 can be assigned to unit D

  In the example of the watermark signal shown in FIG. 14 and FIG. 16, since the number of dots in one unit is all equal, the apparent shading becomes uniform by arranging these units without gaps. Therefore, it appears that a gray image having a single density is embedded as a background on the printed paper.

  In order to produce such an effect, for example, the unit E is defined as a background unit (signal unit to which no symbol is assigned), and these are arranged without gaps as the background of the document image 11, and the symbol units (unit A, unit B) are defined. ) Is embedded in the document image 11, the background unit (unit E) and the symbol unit (unit A, unit B) at the position to be embedded are exchanged.

  FIG. 17A is an explanatory diagram showing a case where the unit E is defined as a background unit and arranged as a background of the document image 11 without gaps. 17 (2) shows an example in which the unit A is embedded in the background image of FIG. 17 (1), and FIG. 17 (3) shows an example in which the unit B is embedded in the background image of FIG. 17 (1). Show. In the present embodiment, a method of using the background unit as the background of the document image 11 will be described. However, only the symbol unit may be arranged as the background of the document image 11.

  Next, a method of embedding one symbol of the code word in the document image 11 will be described with reference to FIG.

  FIG. 18 is an explanatory diagram showing an example of a symbol embedding method in the document image 11. Here, a case where a bit string “0101” is embedded will be described as an example.

  As shown in FIGS. 18A and 18B, the same symbol unit is repeatedly embedded. This is to prevent the character unit in the document from being detected when a signal is detected when it is superimposed on the embedded symbol unit. The symbol unit repetition number and arrangement pattern (hereinafter referred to as a unit pattern) are used. Is optional.

  That is, as an example of the unit pattern, the number of repetitions is set to 4 (four symbol units exist in one unit pattern) as shown in FIG. 18 (1), or the number of repetitions is set to 2 as shown in FIG. 18 (2). (There are two symbol units in one unit pattern) or the number of repetitions may be one (only one symbol unit exists in one unit pattern).

  In FIGS. 18A and 18B, one symbol is given to one symbol unit. However, even if symbols are given to the arrangement pattern of symbol units as shown in FIG. good.

  How many bits of information can be embedded in one page depends on the size of the signal unit, the size of the unit pattern, and the size of the image (or the original image). The number of signals embedded in the horizontal and vertical directions of the image may be detected as known, or may be calculated backward from the size of the image input from the input device and the size of the signal unit.

  If Pw unit patterns in the horizontal direction for one page and Ph unit patterns in the vertical direction can be embedded, unit patterns at arbitrary positions in the image are represented by U (x, y), x = 1 to Pw, y = 1. ~ Ph, and U (x, y) is referred to as a "unit pattern matrix". In addition, the number of bits that can be embedded in one page is referred to as the “number of embedded bits”. The number of embedded bits is Pw × Ph.

  FIG. 19 is a flowchart showing a method for embedding the falsification detection data Data in the document image 11. Here, for example, a case will be described in which the same information is repeatedly embedded in one (one page) document image 11 when converted as a printed matter. By repeatedly embedding the same information, it is possible to extract the embedded information even when the entire unit pattern is filled and the embedded information disappears when the document image 11 and the falsification detection data Data are superimposed. This is to make it possible.

  First, the alteration detection data Data is converted into an N-element code (step S201). This is the same as step S101 in FIG. Hereinafter, the encoded data is referred to as a data code, and the data code expressed by a combination of unit patterns is referred to as a data code unit Du.

  Next, based on the code length of the data code (here, the number of bits) and the number of embedded bits, how many times the data code unit can be embedded in one image is calculated (step S202). In this embodiment, it is assumed that the code length data of the data code is inserted into the first row of the unit pattern matrix. The code length of the data code may be fixed and the code length data may not be embedded.

  The number Dn of embedding the data code unit is calculated by the following equation with the data code length as Cn.

  Here, assuming that the remainder is Rn (Rn = Cn− (Pw × (Ph−1))), the unit pattern matrix is embedded with a data pattern unit of Dn times and a unit pattern corresponding to the first Rn bits of the data code. become. However, the Rn bit of the remainder part does not necessarily have to be embedded.

  In the description of FIG. 20, the unit pattern matrix size is 9 × 11 (11 rows and 9 columns), and the data code length is 12 (numbers 0 to 11 in the figure indicate the codewords of the data code). And

  Next, the code length data is embedded in the first row of the unit pattern matrix (step S203). In the example of FIG. 20, the code length is expressed by 9-bit data and is described as being embedded only once. However, when the unit pattern matrix width Pw is sufficiently large, It can be embedded repeatedly.

  Further, the data code unit is repeatedly embedded in the second and subsequent rows of the unit pattern matrix (step S204). As shown in FIG. 20, data code MSB (most significant bit) or LSB (least significant bit) is embedded in the row direction in order. The example of FIG. 20 shows an example in which the data code unit is embedded seven times and the first 6 bits of the data code are embedded.

  The data embedding method may be embedded so as to be continuous in the row direction as shown in FIG. 20 or may be embedded so as to be continuous in the column direction.

  The superposition of the document image 11 and the falsification detection data Data in the watermark information synthesis unit 15 has been described above.

  As described above, the watermark information synthesis unit 15 superimposes the document image 11 and the falsification detection data Data. The value of each pixel of the watermarked document image is calculated by AND operation (AND) of the corresponding pixel values of the document image 11 and the falsification detection data Data. That is, if either the document image 11 or the falsification detection data Data is 0 (black), the pixel value of the watermarked original image is 0 (black), otherwise 1 (white).

  FIG. 21 is an explanatory diagram showing an example of a watermarked document image. FIG. 22 is an explanatory view showing a part of FIG. 21 in an enlarged manner. Here, the unit pattern shown in FIG. 18A is used. The watermarked document image 11 (output document image 14) is output by, for example, an interface (not shown) of the watermark information synthesis unit 15.

(About the operation of the document input unit 20)
In the following description, it is assumed that the output document image 14 is distributed after being printed on a print medium such as paper, and the input document image 21 is an image obtained by printing a watermarked print document of the output document image 14 with an input device such as a scanner. However, the present invention is not limited to this example, and the output document image 14 may be digital data that is not printed as it is.

  Specifically, as shown in FIG. 23, the output document image 14 is printed on a print medium such as paper and then distributed as a printed material, and the input document image 21 is a watermarked document image printed on a scanner or the like. It is assumed that the image is input by the input device. It is assumed that the input document image 21 shown in FIG. 15 has been tampered with. FIG. 23 is an explanatory diagram illustrating an example of a schematic configuration of the input document image 21.

  First, the watermark information extraction unit 22 provided in the document input unit 20 extracts watermark information (falsification detection data Data) from the input document image 21 and details the falsification detection data Data (falsification detection data Data0, falsification detection). Data Data1, alteration detection data Data2, alteration detection data Data3) are restored. If the falsification detection data Data1 is compressed, decompression is performed to restore the composite image. Further, the falsification detection data Data3 is omissible data, and if omitted, the document output unit 10 and the document input unit 20 perform processing in advance by a common creation method. When the falsification detection data Data2 is quantized, the data includes a quantization error and is restored to the original value.

(About the operation of the watermark information extraction unit 22)
Next, the processing of the watermark information extraction unit 22 will be described with reference to FIG. FIG. 24 is a flowchart illustrating an example of a schematic flow of processing of the watermark information extraction unit 22.

  As shown in FIG. 24, first, a watermarked original image is input to a memory of a computer or the like by an input device (not shown) such as a scanner (step S301). This image is referred to as an input image. The input image is a multi-valued image and will be described below as a gray image with 256 gradations. However, the input image is not limited to this example, and may be a full-color image or the like. Further, the resolution of the input image (resolution when read by an input device such as a scanner (not shown)) may be different from the watermarked document image (output document image 14) created by the document output unit 10. In the following description, it is assumed that the resolution is the same as that of the image created by the document output unit 10. A case where one unit pattern is composed of one symbol unit will be described.

<Signal Detection Filtering Step (Step S310)>
In step S310, a filtering process is performed on the entire input document image 21, and a filter output value is calculated and the filter output value is compared. The filter output value is calculated by convolution between the filter and the image in all pixels of the input document image 21 using a filter called a Gabor filter shown below.

  The Gabor filter G (x, y), x = 0 to gw−1, and y = 0 to gh−1 are shown below. gw and gh are filter sizes, which are the same size as the signal unit embedded by the watermark information embedding apparatus 10 here.

  The filter output value at an arbitrary position in the input image is calculated by convolution between the filter and the image. In the case of a Gabor filter, there are a real number filter and an imaginary number filter (an imaginary number filter is a filter whose phase is shifted by a half wavelength from the real number filter), and the mean square value thereof is used as a filter output value. For example, if the convolution of the luminance value at a certain pixel (x, y) with the real filter of the filter A is Rc and the convolution with the imaginary filter is Ic, the filter output value F (A, x, y) is Calculate with the following formula.

  After calculating the filter output values for all the filters corresponding to each signal unit as described above, the filter output values calculated as described above are compared in each pixel, and the maximum value F (x, y) is obtained. Store as a filter output value matrix. Further, the number of the signal unit corresponding to the filter having the maximum value is stored as a filter type matrix (FIG. 25). Specifically, in a certain pixel (x, y), when F (A, x, y)> F (B, x, y), F is set as the value of (x, y) in the filter output value matrix. (A, x, y) is set, and “0” indicating the signal unit A is set as the value of the filter type matrix (x, y) (in this embodiment, the numbers of the signal units A and B are set to “0”). ”And“ 1 ”).

  In the present embodiment, the number of filters is two. However, when the number of filters is larger than that, the maximum value of a plurality of filter output values and the signal unit number corresponding to the filter at that time are also stored. That's fine.

<Signal Location Search Step (Step S320)>
In step S320, the position of the signal unit is determined using the filter output value matrix obtained in step S310. Specifically, first, assuming that the size of the signal unit is Sh × Sw, a signal in which the vertical interval between grid points is Sh, the horizontal interval is Sw, and the number of grid points is Nh × Nw. A position search template is created (FIG. 26). The size of the template thus created is Th (Sh * Nh) × Tw (Sw * Nw), and optimum values for searching for the signal unit position may be used for Nh and Nw.

  Next, the filter output value matrix is divided for each template size. Further, in each divided region, the template is moved in units of pixels on the filter output value matrix within a range (horizontal direction ± Sw / 2, vertical direction ± Sh / 2) that does not overlap with the signal units in the adjacent region. The sum V of the filter output value matrix values F (x, y) on the lattice points is obtained using the following equation (FIG. 31), and the lattice point of the template having the largest sum is determined as the position of the signal unit in the region. To do.

  The above example is a case where the filter output value is obtained for all the pixels in step S310. When filtering is performed, it is possible to perform filtering only for pixels at a certain interval. For example, when filtering is performed every two pixels, the interval between the lattice points of the signal position search template may be halved.

<Signal Symbol Determination Step (Step S330)>
In step S330, the signal unit is determined to be A or B by referring to the value of the filter type matrix at the signal unit position determined in step S320 (signal unit number corresponding to the filter).

  The determination result of the determined signal unit is stored as a symbol matrix as described above.

<Signal Boundary Determination Step (Step S340)>
In step S320, since the filtering process is performed on the entire image regardless of whether the signal unit is embedded, it is necessary to determine in which part the signal unit is embedded. Therefore, in step S340, the signal boundary is obtained by searching for a pattern determined in advance when embedding the signal unit from the symbol matrix.

  For example, if the signal unit A is necessarily embedded in the boundary where the signal unit is embedded, the number of the signal units A is counted in the horizontal direction of the symbol matrix determined in step S330, and the signal unit A is respectively measured vertically from the center. The position with the largest number of A is defined as the upper end / lower end of the signal boundary. In the example of FIG. 27, since the signal unit A in the symbol matrix is represented by “black” (in terms of value, “0”), the number of signal units A is calculated by counting the number of black pixels in the symbol matrix. Counting can be performed, and the upper / lower end of the signal boundary can be obtained from the frequency distribution. The left end / right end can also be obtained in the same way, except for the direction in which the number of units A is counted.

  In order to obtain the signal boundary, not only the above method but also a pattern that can be searched from the symbol matrix may be determined in advance on the embedding side and the detecting side.

  Returning again to the flowchart of FIG. 24, the following step S305 will be described. In step S305, the original information is restored from the portion corresponding to the inside of the signal boundary in the symbol matrix. In the present embodiment, since one unit pattern is composed of one symbol unit, the unit pattern matrix is equivalent to the symbol matrix.

<Information Decoding Step (Step S305)>
FIG. 28 is an explanatory diagram showing an example of information restoration. The steps of information restoration are as follows.
(1) A symbol embedded in each unit pattern is detected (FIG. 28 (1)).
(2) The symbols are connected to restore the data code (FIG. 28 (2)).
(3) The data code is decoded to extract the embedded information ((3) in FIG. 28).

  29 to 31 are explanatory diagrams showing an example of a data code restoration method. The restoration method is basically the reverse process of FIG.

  First, the code length data portion is extracted from the first row of the unit pattern matrix to obtain the code length of the embedded data code (step S401).

  Next, based on the size of the unit pattern matrix and the code length of the data code obtained in step S401, the number Dn of data code units embedded and the remainder Rn are calculated (step S402).

  Next, data code units are extracted from the second and subsequent rows of the unit pattern matrix by the reverse method of step S203 (step S403). In the example of FIG. 30, 12 pattern units are sequentially decomposed from U (1, 2) (2 rows and 1 column) (U (1, 2) to U (3, 3), U (4, 3) to U (6, 4), ...). Since Dn = 7 and Rn = 6, 12 pattern units (data code units) are extracted 7 times, and 6 unit patterns (corresponding to the upper 6 data code units) (U (4 , 11) to U (9, 11)) are taken out.

  Next, the embedded data code is reconstructed by performing bit certainty calculation on the data code unit extracted in step S403 (step S404). Hereinafter, the bit certainty calculation will be described.

  As shown in FIG. 31, the first outer code units extracted from the second row and the first column of the unit pattern matrix are Du (1,1) to Du (12,1), and Du (1,2) to Du are sequentially added. (12, 2),. Further, the surplus portion is assumed to be Du (1, 8) to Du (6, 8). The bit certainty calculation is to determine the value of each symbol of the data code by, for example, taking a majority vote for each element of each data code unit. As a result, for example, even if signal detection cannot be performed correctly from any unit in any data encoding unit due to overlap with the character area or dirt on the paper surface, it will eventually be correctly The data code can be restored.

  Specifically, for example, the first bit of the data code is more often when the signal detection result of Du (1, 1), Du (1, 2),..., Du (1, 8) is 1. Is determined to be 1 and when there are more 0s, it is determined to be 0. Similarly, the second bit of the data code is determined by majority decision based on the signal detection result of Du (2, 1), Du (2, 2),..., Du (2, 8), and the 12th bit of the data code is Du (12,1), Du (12,2),..., Du (12,7) (Du (12,8) is not present, so it is up to Du (12,7)). Judgment by majority vote.

  Although the case where data codes are repeatedly embedded has been described here, a method that does not repeat data code units can be realized by using an error correction code or the like when encoding data.

  As described above, since the filtering process is performed on the entire input document image 21 and the position of the signal unit can be obtained by using the signal position search template so that the total sum of the filter output values is maximized. Even when the image expands or contracts due to distortion or the like, the position of the signal unit can be detected correctly, and the falsification detection data Data can be detected accurately from the watermarked original image (input document image 21).

(Processing of the pseudo document image generation unit 23)
Next, processing of the pseudo document image generation unit 23 according to the present embodiment will be described with reference to FIGS. 32 and 33. FIG. 32 is a flowchart showing an outline of processing of the pseudo document image generation unit 23. FIG. 33 is an explanatory diagram showing an outline of a pseudo document image generation process by the pseudo document image generation unit 23.

  Based on the falsification detection data Data extracted by the watermark information extraction unit 22 by the processing of the pseudo document image generation unit 23, an image (pseudo document image) in which the image features of the document image 11 are reproduced from the falsification detection data Data is created. To do.

(About reduced character area separation (S41))
As shown in FIG. 32, the pseudo document image generation unit 23 performs the process of reversing each reduced character from the synthesized image by the reverse processing procedure of the reduced document area synthesis (S33) by the watermark information synthesis unit 13 provided in the document output unit 10 described above. The areas are separated (S41). As shown in FIG. 33, the process corresponding to the reduced character area separation process (S41) shown in FIG. 32 is S61.

(About reduced character area enlargement (S42))
When the reduced document area is separated, the pseudo document image generation unit 23 enlarges the reduced reduced character area (S42). The enlargement process (S42) enlarges each reduced character area and restores the original character area size. For example, if the character area has been reduced to ½, after the reduced character area is separated, the reduced character area is enlarged twice. As shown in FIG. 33, the process corresponding to the enlargement process (S42) shown in FIG. 32 is S62.

(Character area reconstruction (S43))
When the reduced character area after the separation is enlarged (S42), the pseudo document image generation unit 23 generates a canvas image (background image) having only the background (white image area) having the same size as the document image 11. .

  Further, the pseudo document image generation unit 23 refers to the falsification detection data Data2 (position information of each character area) and arranges the character area obtained by enlarging the reduced character area on the canvas image (S43). As shown in FIG. 33, the process corresponding to the character area reconstruction process (S43) shown in FIG. 32 is S63.

  Although the canvas image and the pseudo document image have substantially the same size, they do not necessarily have to be the same as the document image 11. If the canvas image and the pseudo document image are not the same as the document image 11, the enlargement ratio is changed in the enlargement process (S42) of the reduced character area. For example, if the width of the canvas image is 1 / Sw of the document image 11 and the height is 1 / Sh, and the character area is reduced to 1 / N, the width is (1 / Sw) / (1 / N) times and height are obtained as (1 / Sh) / (1 / N) times, respectively.

  As described above, the pseudo document image generation unit 23 completes the pseudo document image when all the character regions after the enlargement process (S42) are arranged on the canvas image based on the falsification detection data Data2 (S43).

(Regarding the processing of the input image transformation unit 24)
When the pseudo document image is generated by the pseudo document image generation unit 23, the input document image 21 is deformed by the input image deformation unit 24. In the transformation process, the input document image 21 is corrected to an image having the same size as that of the pseudo document image to be compared for performing alteration detection, and the corrected input document image 21 is binarized. In addition, after correcting to the same size as the pseudo document image and binarization, the image is reduced by the same rate as the reduction rate of the reduction process (S32) in order to improve the accuracy of falsification detection, and further the enlargement rate of the enlargement process (S42) Magnify at the same rate as. In other words, by using the same conditions as the processing that has been performed until the pseudo document image is generated, the accuracy of the determination of whether or not falsification has occurred can be further improved, but it can be omitted.

  Here, the processing of the input image transformation unit 24 will be described with reference to FIG. FIG. 34 is a flowchart of the input image deformation unit 24. Hereinafter, description will be given according to this flowchart.

<Signal Unit Position Detection (Step S610)>
FIG. 35 shows the signal unit position detected in the first embodiment on the input image (original image 11 with watermark) 21. In FIG. 35, the signal unit is expressed as U (x, y), x = 1 to Wu, y = 1 to Hu. U (1, y) to U (Wu, y) are signal units (reference numeral 710 in FIG. 35), and U (x, 1) to U (x, Hu) are signal units (in the same column). Reference numeral 720 in FIG. U (1, y) to U (Wu, y), U (x, 1) to U (x, Hu), etc. are not actually arranged on the same straight line, and are slightly shifted vertically and horizontally. Yes.

  The coordinate value P on the input image of the signal unit U (x, y) is expressed as (Px (x, y), Py (x, y)), x = 1 to Wu, y = 1 to Hu (FIG. 35, 730, 740, 750, 760). However, the input image is filtered every N pixels vertically and horizontally (N is a natural number). This filtering is performed in the same manner as in the <signal position search step (step S320)> in the first embodiment. P is a value obtained by simply multiplying the coordinate value of each signal unit in the signal output value matrix N times vertically and horizontally.

<Linear approximation of signal unit position (step S620)>
Approximate the position of the signal unit with a straight line in the row and column directions. FIG. 36 shows an example of linear approximation in the row direction. In FIG. 36, the positions of signal units U (1, y) to U (Wu, y) in the same row are approximated by a straight line Lh (y). The approximate straight line is a straight line that minimizes the sum of the distances between the position of each signal unit and the straight line Lh (y). Such a straight line can be obtained by a general method such as a least square method or principal component analysis. The straight line approximation in the row direction is performed for all the rows, and the straight line approximation in the column direction is similarly performed for all the columns.

  FIG. 37 shows an example of a result obtained by performing linear approximation in the row direction and the column direction. In FIG. 37, the signal unit is represented as U (x, y), x = 1 to Wu, y = 1 to Hu. Lh (y) is a straight line (reference numeral 810 in FIG. 37) approximating U (1, y) to U (Wu, y), and Lv (x) is U (x, 1) to U (x, Hu). Is a straight line (reference numeral 820 in FIG. 37).

<Straight line equalization (step S630)>
The straight lines approximated in step S620 are not uniform in slope or position when viewed individually, for example, because the positions of the detected signal units are shifted to some extent. Therefore, in step S630, equalization is performed by correcting the inclination and position of each straight line.

  FIG. 38 shows an example of correcting the inclination of the approximate straight line Lh (y) in the row direction. FIG. 38A shows a state before correction, and FIG. 38B shows a state after correction. When the slope of Lh (y) in FIG. 38A is Th (y), the slope of Lh (y) is corrected so as to be the average value of the slopes of straight lines in the vicinity of Lh (y). Specifically, Th (y) = AVERAGE (Th (y−Nh) to Th (y + Nh)). However, AVERAGE (A to B) is a calculation formula for calculating an average value from A to B, and Nh is an arbitrary natural number. When y−Nh <1, Th (y) = AVERAGE (Th (1) to Th (y + Nh)), and when y + Nh> Hu, Th (y) = AVERAGE (Th (y−Nh) to Th (Hu). )). FIG. 38 shows an example where Nh is 1, and FIG. 38B shows an example where Lh (y) is corrected by the average value of the slopes of the straight lines Lh (y−1) to Lh (y + 1). Show.

  FIG. 39 shows an example of correcting the position of the approximate straight line Lh (y) in the row direction. FIG. 39A is before correction, and FIG. 39B is after correction. In FIG. 39A, when an arbitrary reference straight line 1130 is set in the vertical direction and the y coordinate of the intersection of this straight line and Lh (y) is Q (y), Q (y) becomes Lh (y). It correct | amends so that it may become the average of the linear position of near. Specifically, Q (y) = AVERAGE (Q (y−Mh) to Q (y + Mh)). However, Mh is an arbitrary natural number. If y-Mh <1 or y + Mh> Hu, no change is made. FIG. 39 shows an example when Mh is 1, and FIG. 39B shows that Lh (y) is corrected by the midpoint (average) of the positions of the straight lines Lh (y−1) and Lh (y + 1). An example is shown. This process can be omitted.

<Line intersection calculation (step S640)>
Calculate the intersection of the approximate line in the row direction and the approximate line in the column direction. FIG. 40 shows an example in which the intersection points of the approximate straight lines Lh (1) to Lh (Hu) in the row direction and the approximate straight lines Lv (1) to Lv (Wu) in the column direction are calculated. The intersection point is calculated by a general mathematical method. The intersection calculated here is used as the corrected signal unit position. That is, the intersection of the approximate straight line Lh (y) in the row direction and the approximate straight line Lv (x) in the column direction is set to the corrected position (Rx (x, y), Ry (x, y) of the signal unit U (x, y). y)). For example, the corrected position of the signal unit U (1, 1) is an intersection of Lh (1) and Lv (1).

<Creation of corrected image (step S650)>
A corrected image is created from the input image with reference to the signal unit position calculated in step S640. Here, the resolution when printing the watermarked original image output by the watermark image output unit 10 is Dout, and the resolution when the input document image 21 to be input to the document input unit 20 is acquired is Din. The corrected input document image 21 is assumed to be the same size as the pseudo document image.

  Assuming that the size of the signal unit in the watermark image output unit 10 is the width Sw and the height Sh, the signal unit in the input document image 21 has a width Tw = Sw × Din / Dout, and the height is Th = Sh × Din / Dout. It becomes. Therefore, when the number of signal units is Wu in the horizontal direction and Hu in the vertical direction, the size of the input document image 21 after correction is Wm = Tw × Wu and Hm = Th × Hu. . Further, if the position of an arbitrary signal unit U (x, y) in the corrected input document image 21 is (Sx (x, y), Sy (x, y)), correction is performed so that the signal units are evenly arranged. In order to create a later input document image 21, Sx = Tw × x and Sy = Th × y hold. The position of the upper left signal unit U (1, 1) is (0, 0), which is the origin of the input document image 21 after correction.

  The pixel value Vm at an arbitrary position (Xm, Ym) on the corrected image is obtained from the pixel value Vi at the coordinates (Xi, Yi) on the input document image. FIG. 41 is a correspondence example of these coordinates. FIG. 41A shows the input image 1310 (input document image 21), and FIG. 41B shows the corrected image 1320 (corrected input document image 21). Yes. The relationship between (Xm, Ym) and (Xi, Yi) will be described using this figure.

  In the corrected image 1320 of FIG. 41 (b), the closest signal units in the upper left, upper right, and lower left areas when viewed at (Xm, Ym) as the center are U (x, y) (coordinate values are (Sx ( x, y), Sy (x, y)), 1360), U (x + 1, y) (1370), U (x, y + 1) (1380), and their distances as E1, E2, E3, respectively. (Specifically, x is the smallest integer not exceeding Xm / Tw + 1, and y is the smallest integer not exceeding Xm / Tw + 1). At this time, U (x, y) (coordinate values are (Rx (x, y), Ry (x, y)), 1330), U (x + 1, y) (1340) in the input image 1310 in FIG. ), U (x, y + 1) (1350) and (Xi, Yi) are D1, D2, and D3, respectively, and the ratio D1: D2: D3 of D1 to D3 is equal to E1: E2: E3. The pixel value Vm of (Xm, Ym) is obtained from the pixel value Vi of the coordinates (Xi, Yi) on the input image 1310.

  FIG. 42 shows a specific calculation method of such (Xi, Yi). Reference numeral 1430 in FIG. 42A is a point obtained by projecting (Xm, Ym) onto a straight line connecting U (x, y) and U (x + 1, y), and Fx = Xm−Sx (x, y). Reference numeral 1440 denotes Fy = Ym−Sy (x, y) at a point where (Xm, Ym) is projected onto a straight line connecting U (x, y) and U (x, y + 1). Similarly, reference numeral 1450 in FIG. 42B is a point where (Xm, Ym) is projected onto a straight line connecting U (x, y) and U (x + 1, y). Gx = Xm−Sx (x, y) It is. Similarly, reference numeral 1460 is Gy = Ym−Sy (x, y) at a point where (Xm, Ym) is projected onto a straight line connecting U (x, y) and U (x, y + 1). At this time, Fx in the input image 1410 in FIG. 42A is Fx / (Rx (x + 1, y) −Rx (x, y)) = Ex / Tw, Fx = Ex / Tw × (Rx (x + 1, y) -Rx (x, y)). Similarly, Fy = Ey / Th × (Ry (x, y + 1) −Ry (x, y)). Thus, Xi = Fx + Rx (x, y) and Yi = Fy + Ry (x, y).

  As described above, the pixel value of an arbitrary point (Xm, Ym) on the corrected image 1420 (corrected input document image 21) in FIG. 42B is set to the point (Xi) on the input image (input document image 21). , Yi) is set. However, since (Xi, Yi) is generally a real value, the pixel value at the coordinates closest to (Xi, Yi) on the input image is set, or the pixel values of the four neighboring pixels and the distance between them are set. Calculate the pixel value from the ratio.

  The operation of the input image deforming unit 24 has been described above. The input image deforming unit 24 further performs a correction and binarization process on the input document image 21, and then reduces the input document image 21 after the correction. On the contrary, a process of performing an enlargement process by the reduced amount is executed.

  Here, the input document image 21 corrected and binarized by the input image deforming unit 24 and further reduced / enlarged is shown in FIG. FIG. 43 is an explanatory diagram showing an example of a schematic configuration of an input document image when the corrected and binarized input document image is further reduced / enlarged.

  As described above, according to the present embodiment, an image captured from a printed material is used to correct an image captured from a printed document based on position information (falsification detection data Data2) of a signal embedded in the input document image 21. Therefore, the image before printing can be restored without distortion or expansion / contraction, so that the positions of the images can be correlated with high accuracy, and further, high-performance alteration detection can be performed.

(About the processing of the falsification detection unit 25)
When the correction process of the input document image 21 is completed by the input image deformation unit 24, the falsification detection unit 25 compares the input document image 21 after the correction and the pseudo document image with each other to thereby compare the input document image 21. It is determined whether or not tampering exists in 21.

  Here, the processing of the falsification detection unit 25 according to the first embodiment will be described with reference to FIG. FIG. 44 is a flowchart showing an outline of processing of the falsification detection unit according to the first embodiment.

  As shown in FIG. 44, first, the falsification detection unit 25 obtains a difference between the pseudo document image and the input document image 21 after the correction and reduction / enlargement processing, and generates a difference image (S51).

  The difference image generated by the falsification detection unit 25 is illustrated as shown in FIG. The difference image shown in FIG. 45 is an example of an image generated by the difference between the input document image 21 reduced and enlarged after the correction shown in FIG. 43 and the pseudo document image. The black area in the image is the difference. For example, it can be seen that there are many black regions near the outline of the character.

  The difference image is a difference between the pseudo document image and the corrected / reduced / enlarged input document image 21. In addition to the case where a black region (difference region) exists near the outline of the character, for example, the entire character Is a black area, and in the above case, it is determined that the black area is falsified. However, even in the above-described case, there is a difference area that determines that a part that has not been falsified has been falsified because the binarization threshold of the input image deformation unit 24 is not optimal. May appear.

Therefore, in order to avoid the case where it is erroneously determined to be falsified as described above, the falsification detection unit 25 performs noise removal processing (S52) on the difference image. The noise removal process, for example, the difference area of the difference image to perform the labeling process, if the same label with a (grouped) by the sum of the number of pixels in the difference region is less than the threshold value T _n, the grouping The determined difference area is determined to be noise, and the difference area is removed from the difference image. As a means for removing the difference area, for example, a black pixel can be changed to a white pixel. Such noise removal can prevent misjudgment, for example, when a black region is present in the outline of a character without being altered.

  After the noise removal process (S52), the alteration detection unit 25 determines that the area remaining as the difference area has been altered (S53).

  As shown in FIG. 45 and FIG. 46, the difference area determined to be falsified by the falsification detection unit 25 is the portion of “11” described in the “student ID number” column and “psychology” “ It can be seen that there are two places where “A” is written in the “Results” column. That is, the above-described “11” and “A” are the falsified areas (difference areas) from which they are detected.

  This is the end of the description of the series of operations of the document input unit 20. According to the first embodiment, not the entire image area of the document image 11 but only the character area is cut out, and an image in which the character areas are collectively collected is embedded in the document image 11 as watermark information. The data amount of the falsification detection data Data (watermark information) for detection can be reduced. Therefore, the processing for embedding the falsification detection data Data in the document image can be performed quickly and efficiently, and the processing for extracting the embedded falsification detection data Data can be performed quickly and efficiently, and the amount of watermark information can be reduced. High-precision alteration detection can be automatically executed.

(About the second embodiment)
In the first embodiment, a case has been described in which character areas existing in the document image 11 are reduced at the same reduction ratio. In the second embodiment, a case will be described in which it is assumed that there are a plurality of character areas in the document image 111 and that the importance of the description content and the like is different for each character area.

  In the second embodiment, for example, a character area of a description item with high importance is reduced to determine whether or not there is a finer alteration by reducing the reduction rate, and a character area of description items with low importance is reduced. Increase the rate to determine whether there has been a rough tampering.

  Next, the document output unit 110 provided in the falsification detection apparatus according to the second embodiment will be described with reference to FIG. FIG. 47 is a block diagram illustrating an example of a schematic configuration of a document output unit according to the second embodiment. Note that the falsification detection device includes a document output unit 110 and / or a document input unit 120.

  Further, the description of the document output unit 110 provided in the falsification detection apparatus according to the second embodiment will be described in detail with respect to differences from the document output unit 10 according to the first embodiment. About a point, since it is substantially the same structure, detailed description is abbreviate | omitted.

  As shown in FIG. 47, the difference between the document output unit 110 according to the second embodiment and the document output unit 10 according to the first embodiment is the reduction ratio determination unit 115 in the document output unit 110. Is added, instead of the document feature data conversion unit 12 included in the document output unit 10, a second document feature data conversion unit (image processing unit) 116 is added instead.

  The reduction rate determination unit 115 refers to the importance of each character area and determines a reduction rate when reducing the character area to a reduced character image.

  Next, the document input unit 120 provided in the falsification detection apparatus according to the second embodiment will be described with reference to FIG. FIG. 48 is a block diagram illustrating an example of a schematic configuration of a document input unit according to the second embodiment. The tampering detection apparatus includes at least one of the document output unit 110 and the document input unit 120.

  Note that the description of the document input unit 120 provided in the falsification detection apparatus according to the second embodiment will be described in detail with respect to differences from the document input unit 20 according to the first embodiment. About a point, since it is substantially the same structure, detailed description is abbreviate | omitted.

  As shown in FIG. 48, the difference between the document input unit 120 according to the second embodiment and the document input unit 20 according to the first embodiment is the pseudo document image generation provided in the document input unit 20. Instead, the second pseudo document image generating unit 127 is provided instead of the unit 23, and the second input image deforming unit (image deforming unit) 228 is provided in the document input unit 120 instead of the input image deforming unit 24. It is.

(Operation of the second document feature data conversion unit 116)
Next, a series of operations of the second document feature data converting unit 116 will be described with reference to FIG. FIG. 49 is a flowchart showing an outline of processing of the second document feature data converting unit 116 according to the second embodiment.

  As shown in FIG. 49, first, the second document feature data conversion unit 116 extracts a character area from the document image 111 (S131). The character area extraction process (S131) for extracting the character area according to the second embodiment is substantially the same as the character area extraction process (S31) according to the first embodiment, and thus detailed description will be made. Is omitted.

  Next, the second document feature data conversion unit 116 refers to the importance level for each description item block set in advance in the document image 111, and stores each character area so as to correspond to the importance level of the description item block. A reduction ratio is set (S132).

  Here, with reference to FIG. 50, the importance of each description item block according to the second embodiment will be described. FIG. 50 is an explanatory diagram illustrating an example of a schematic configuration of importance levels of the description item blocks according to the second embodiment.

  As shown in FIG. 50, one or two or more description item blocks 300 (the description item block 300a to the description item block 300c) are set in the document image 111. The entry block 300 is set in advance with the position / size of the entry block 300 before embedding watermark information by a user or the like who creates the document image 111. The degree is set.

  In the document image 111 shown in FIG. 50, a description item block 300a, a description item block 300b, and a description item block 300c are set. Further, regarding the importance, the importance of the entry block 300a and the entry block 300c is set to “medium”, and the importance of the entry block 300b is set to “high”. It should be noted that it is not necessary to detect falsification for an area in which the description item block 300 is not set because the importance is low.

  The importance according to the second embodiment has been described by taking the case of “medium” or “high” as an example, but is not limited to such an example. For example, the importance is “medium” Even if the importance increases in the order of "", "high", "highest", etc., it can be implemented.

  In the document image 111 shown in FIG. 50, among the extracted character areas (S131), the character areas belonging to the entry block 300a and the entry block 300c are increased in reduction ratio (for example, reduced to ¼). ), The reduction ratio is reduced (for example, reduced to 1/2) for the character area belonging to the description item 300b.

  The importance of the entry block 300 according to the second embodiment has been described by taking as an example the case where the importance is set in advance before embedding as watermark information. However, the importance is not limited to this example. For example, each time a character area existing in the entry block 300 is extracted, the importance is recognized, and the keyword DB (database) is provided in the falsification detection device, so that the keyword stored in the keyword DB This can be implemented even when the character existing in the character area matches and the importance is dynamically set to “high” or “highest”.

  When the second document feature data conversion unit 116 executes the character region extraction process (S131) and further refers to the importance set for each description item block 300, the second document feature data conversion unit 116 Assign region numbers to regions.

  FIG. 51 shows an example of the result of assigning numbers to character areas by the second document feature data converting unit 116. FIG. 51 is an explanatory diagram showing an example of the result of the number allocation process by the second document feature data converting unit 116 according to the second embodiment.

  As shown in FIG. 51, for example, a description item block 300 is set as a target to which a number is assigned to a character region, and a character region belonging to the description item block 300 whose importance level is “medium” or higher. An area number is assigned to each, but the present invention is not limited to such an example.

  As shown in FIG. 51, as the priority order in which numbers are assigned to character areas, first, the number of the description item block 300 is the first priority (the number of the description item block 300 is the lower priority). To do. Therefore, as shown in FIGS. 50 and 51, numbers are assigned in order from the character area belonging to the entry block 300a, then numbers are assigned to the character areas belonging to the entry block 300b, and finally the entry area 300c. Assign a number to the character area.

  As the next priority, the y-coordinate value of the upper left vertex of the rectangle of the character area is given the second priority. It is assumed that the upper left vertex of the rectangle of the document image 111 is the reference coordinate (0, 0). Accordingly, among the character areas belonging to the entry block 300, numbers are assigned in ascending order from the character area in which the y-coordinate value of the upper left vertex of the rectangle is close to the reference coordinates (0, 0).

  Finally, as a priority order in which numbers are assigned to the character area, the x coordinate value is the third priority. Therefore, when the y-coordinate value, which is the second priority among the character areas belonging to the entry block 300, is the same value, the x-coordinate values are assigned in ascending order from the closest to the reference coordinates (0, 0).

  As described above, when the second document feature data conversion unit 116 assigns numbers to the character areas belonging to the description item block 300 in accordance with the number assignment priority order, the result is as shown in FIG.

  As shown in FIG. 51, since the areas 2 to 5 which are character areas belong to the description item block 300a, the second document feature data conversion unit 116 sets the reduction ratio of the areas 2 to 5 to “4” ( Since the areas 8 to 9 belong to the description item block 300c, the second document feature data conversion unit 116 sets the reduction ratio of the areas 8 to 9 to “4” (1 / 4) and finally, since the areas 10 to 11 belong to the description item block 300b, the second document feature data conversion unit 116 sets the reduction ratio of the areas 10 to 11 to “2” (1 / Set to 2).

  When the reduction ratio for each character area is determined (S132), the second document feature data converting unit 116 cuts out the character area from the document image 111, and reduces the character area at a set reduction ratio corresponding to the character area. (S132).

  With reference to FIG. 52, the character region segmentation processing and the character region reduction processing by the second document feature data converting unit 116 will be specifically described. FIG. 52 is an explanatory diagram showing an example of an outline of the character region cutout and reduction processing (S132).

  As shown in FIG. 52, since the area 2 belonging to the description item block 300a has a reduction ratio of “4”, the second document feature data converting unit 116 extracts the area 2 after cutting out the area 2. It is reduced to 1/4 (S132). The other areas (area 3 to area 5) belonging to the entry block 300a are processed in the same manner as the area 2.

  As shown in FIG. 52, since the reduction ratios of the area 10 and the area 11 belonging to the description item block 300b are “2”, after the area 10 and the area 11 are cut out, the second document feature data is generated. The unit 116 reduces the areas 10 and 11 to ½ (S132). The reduced character area is defined as a reduced character area.

  Next, the second document feature data converting unit 116 arranges the reduced character areas so as to combine the reduced character areas, and generates one image (composite image). Note that the process for generating a composite image according to the first embodiment (S34) and the process for generating a composite image according to the second embodiment (S134) are substantially the same, so a detailed description of the process will be given. Omitted.

  Here, a synthesized image according to the second embodiment will be described with reference to FIG. FIG. 53 is an explanatory diagram illustrating an example of a schematic configuration of a composite image according to the second embodiment.

  As shown in FIG. 53, in the composite image, one or two or more reduced character regions having different reduction ratios are embedded without any gaps. Note that the composite image shown in FIG. 53 is generated by arranging the reduced character areas in the vertical direction (vertical direction) in ascending order of the area numbers, similarly to the composite image shown in FIG. The difference between the composite image shown in FIG. 10 and the composite image shown in FIG. 53 is that the composite image shown in FIG. 53 is mixed with reduced character areas having different reduction ratios. Others are substantially the same as the synthesized image shown in FIG.

(Regarding the processing of the watermark information synthesis unit 113)
A watermark information synthesis unit (watermark information embedding unit) 113 embeds data for falsification detection (falsification detection data Data) as watermark information in the document image 111. Note that the falsification detection data Data according to the second embodiment is assumed to be the following data.

Data 100: Header information such as the size and data size of the document image 111 Data 101: Image data of the composite image or image data obtained by compressing the composite image Data 102: Position information of the character area (extracted by the character area extraction process (S131)) And reduction rate (or reduction level)
Data 103: Composite image creation method ID (when the composite image creation method is changed for each document image 111)

  The falsification detection data Data102 can be exemplified by tabulated data as shown in FIG. FIG. 54 is an explanatory diagram illustrating a schematic configuration of the falsification detection data Data 102 according to the second embodiment.

  As shown in FIG. 54, for example, in the case of a character area whose “area number” is “1” and whose “reduction level” is “0”, as described above, the character area belonging to the entry block 300 with low importance Therefore, the image data of the character area is not recorded in the falsification detection data Data101.

  On the other hand, as shown in FIG. 54, for example, in the case of a character area whose “area number” is “N” and whose “reduction level” is “1” (for example, the reduction ratio is “2”, etc.) Therefore, the image data of the character area is recorded in the falsification detection data Data101.

Similarly, as shown in FIG. 54, for example, “area number” is “2”, and “reduction level” is “2”.
In the case of a character area (for example, a reduction ratio of “4” or the like), since it is a character area belonging to the entry block 300 having a medium importance level, the falsification detection data Data101ni image data of the character area is recorded.

  If the correspondence between the importance level and the reduction ratio of the description item block 300 is common to both the document output unit 110 and the document input unit 120, the falsification detection data Data102 is applied to the first embodiment. It is also possible to change to the same data structure as the falsification detection data Data2 and to add the falsification detection data Data104 shown below.

  Data 104: Information such as the position / size of the entry block 300 and the importance level

  Here, the falsification detection data Data104 will be described with reference to FIG. FIG. 55 is an explanatory diagram showing an example of a schematic configuration of the falsification detection data Data 104 according to the second embodiment.

  As shown in FIG. 55, the falsification detection data Data 104 includes “description item” indicating the number of the descriptive item block 300, “Left”, “Top”, and “Right”, which are information regarding the position of the descriptive item block 300. And “Bottom”, and “Importance” for indicating the importance of the character area belonging to the entry block 300.

  Note that processing such as embedding the falsification detection data Data into the document image 111 by the watermark information combining unit 113 according to the second embodiment is almost the same as the processing by the watermark information combining unit 13 according to the first embodiment. Since it is the same structure, detailed description is abbreviate | omitted.

(About the operation of the document input unit 120)
In the following description, it is assumed that the output document image 114 is distributed after being printed on a print medium such as paper, and the input document image 121 is an image obtained by printing a watermarked print document of the output document image 114 with an input device such as a scanner. However, the present invention is not limited to this example, and the output document image 114 may be digital data that is not printed as it is.

  A case will be described in which the falsification detection data Data102 is used as the falsification detection data Data embedded in the document image 111 in the document output unit 110 according to the second embodiment.

  The processing of the document input unit 120 according to the second embodiment will be described with respect to differences from the processing of the document input unit 20 according to the first embodiment, and the other points will be substantially the same. Since this is a simple configuration, it is omitted.

(Processing of the second pseudo document image generation unit 127)
The operation of the second pseudo document image unit 127 is substantially the same as the operation of the pseudo document image unit 23 according to the first embodiment. However, in the second embodiment, the reduction rate is different for each character area. Since they are different, the enlargement ratio is changed for each character area accordingly. Note that image data of a character area that is not recorded in the falsification detection data Data 101 as a composite image is not reflected in the pseudo document image.

  As shown in FIG. 56, the pseudo document image generated by the second pseudo document image unit 127 is not necessarily completely coincident with the document image 111, and as indicated by an arrow, characters not recorded in the falsification detection data Data 101 are displayed. Since the area is low in importance, such a character area is not arranged in the pseudo document image in the first place.

  Therefore, for example, if the character area arranged in the pseudo document image has an importance level of “medium” or higher, the pseudo document image is generated by being arranged by the second pseudo document image generation unit 127.

(About operation | movement of the 2nd input image deformation | transformation part 128)
As described in the input image transformation unit 24 according to the first embodiment, the second input image transformation unit 128 applies a character area included in the falsification detection data Data 102 to the binarized input document image 121. After referring to the coordinate data indicating the position of the character and cutting out the character area corresponding to the position / size, the character area is temporarily reduced at the reduction ratio with reference to the “reduction level” included in the alteration detection data Data102. , After enlarging with the reciprocal of the reduction ratio, the binarized input document image 121 is placed at the original position.

  Here, the operation of the second input image transformation unit 128 will be described with reference to FIG. FIG. 57 is an explanatory diagram showing an outline of the operation of the second input image deformation unit 128.

  As shown in FIG. 57, the second input image transformation unit 128 cuts out the character area of area 2 in the input document image 121 according to the position information (coordinate values) of the falsification detection data Data102 (S131). The other character areas are processed in the same manner as described above.

  Next, the second input image transformation unit 128 refers to the “reduction level” of the falsification detection data Data 102 corresponding to the cut-out area 2 and the reduction level is “2”. The area is reduced to ½ (S132). The other character areas are processed in the same manner as described above.

  Next, the second input image deforming unit 128 enlarges the character area in the area 2 reduced in S132 by “2” times the reciprocal of the reduction ratio “1/2” (S133).

  Next, the second input image transformation unit 128 arranges the character area of the area 2 enlarged by the enlargement process at the original location of the input document image 121 (S134). It should be noted that for the character area having “0” as the “reduction level” included in the alteration detection data Data 102, the second input image transformation unit 128 fills the character area with the background color (white in the case of FIG. 57).

  In S134, the second input image transformation unit 128 has been described by taking as an example the case where the character area with the “reduction level” of “0” included in the falsification detection data Data102 is filled with the background color. For example, the second input image deforming unit 128 generates a background image (for example, an image composed of white pixels) having the same size as the input document image 121 in advance, and the background image includes the above-described background image. The present invention can be implemented even when the character area of the area 2 enlarged by the enlargement process is pasted at the same position as the original location of the input document image 121.

  When the falsification detection data Data 104 generated by the document output unit 110 is used, the second input image transformation unit 128 has the falsification detection data Data 104 in which position information (coordinate values) of each character area included in the falsification detection data Data2 is changed. The description item block 300 is determined, and the importance level of the description item block 300 is referred to, and the reduction rate and the enlargement rate are determined.

  The reason why the character area of the input document image 121 is separately reduced and enlarged by the second input image deformation unit 128 is to reproduce the state in which each character area of the pseudo document image according to the second embodiment is deformed. This is to improve the performance of tamper detection. The deformed state can be exemplified by, for example, how the characters are crushed.

  For example, if the reduction ratio of each character area is changed according to the importance level, it is possible to inspect whether the important character area has been tampered with in detail. By performing the falsification detection process, the falsification detection process can be speeded up. Specifically, an image obtained by doubling the input document image 121 reduced to ½ is smaller in image deterioration than an image obtained by magnifying the input document image 121 reduced by ¼ to four times. , Detailed alteration detection processing can be executed.

  From the above, in the falsification detection processing according to the second embodiment, since the reduction ratio is changed for each character area, the detection accuracy for detecting whether or not the document has been falsified is determined according to the importance of the contents described in the document. It can be changed dynamically or statically.

(About the third embodiment)
In the first embodiment and the second embodiment, the falsification detection process when the document image (original image) 11 and the document image (original image) 111 include only a character area in which the document is described. explained. In the third embodiment, falsification detection processing when a document image (original image) 211 contains a document and ruled lines will be described.

  When an image region including ruled lines exists in the document image, if the ruled lines are simply processed as image data, the amount of data embedded in the document image as watermark information increases. In addition, if the ruled line is not image data but position information indicating the position of the ruled line is embedded and managed in the document image using coordinate values or the like, the line type of the ruled line such as a double line or a broken line needs to be recorded in the position information. However, automatically determining the line type from the document image requires complicated processing, which increases the processing time for generating watermark information.

  In the first place, it is considered that the ruled line is very unlikely to correspond to confidential information, and it is very rare that a minute change to the ruled line corresponds to tampering. Therefore, in the third embodiment, on the premise of the reason for the ruled lines, the ruled lines are removed, the watermark information is embedded in the document image, and the ruled lines are also removed when the watermark information is extracted to detect falsification. Do tamper detection above. This will be described in detail below.

  The ruled lines according to the third embodiment will be described by taking the case of a black straight line as an example. However, the ruled line is not limited to this example. For example, the ruled line may be red, green, or the like other than black. Even in the case of a broken line, it can be implemented.

  Next, the document output unit 210 provided in the falsification detection apparatus according to the third embodiment will be described with reference to FIG. FIG. 58 is a block diagram illustrating a schematic configuration of a document output unit according to the third embodiment. The tampering detection apparatus includes a document output unit 210 and / or a document input unit 220.

  Further, the description of the document output unit 210 provided in the falsification detection apparatus according to the third embodiment will be described in detail with respect to differences from the document output unit 10 according to the first embodiment. About a point, since it is substantially the same structure, detailed description is abbreviate | omitted.

  As shown in FIG. 58, the document output unit 210 according to the third embodiment has a ruled line processing unit (straight line / broken line processing unit, straight line / line) as compared with the document output unit 10 according to the first embodiment. This is different in that a broken line removal unit 215 is further added. The ruled line processing unit 215 will be described in detail later.

  Next, a document input unit 220 according to the third embodiment will be described with reference to FIG. FIG. 59 is a block diagram illustrating a schematic configuration of the document input unit 220 according to the third embodiment.

  Note that the description of the document input unit 220 according to the third embodiment will be described in detail with respect to differences from the document input unit 20 according to the first embodiment. Detailed description is omitted because of the same configuration.

  As illustrated in FIG. 59, the document input unit 220 according to the third embodiment has a third input instead of the input image deformation unit 24 as compared with the document input unit 20 according to the first embodiment. The difference is that an image deformation unit (image deformation unit) 227 is provided. The third input image deformation unit 227 will be described in detail later.

(About the operation of the document output unit 210)
Next, the operation of the ruled line processing unit 215 provided in the document output unit 210 according to the third embodiment will be described with reference to FIG. FIG. 60 is a flowchart illustrating an outline of processing of the ruled line processing unit 215 according to the third embodiment. The operation of the document output unit 210 according to the third embodiment will be described with respect to differences from the document output unit according to the first or second embodiment.

  As shown in FIG. 60, the ruled line processing unit 215 first executes a ruled line extraction process (S231) for extracting a ruled line from the document image 211. Next, the ruled line processing unit 215 executes ruled line removal processing (S232) for removing the ruled lines extracted in S231.

  Here, a document image 211 according to the third embodiment will be described with reference to FIG. FIG. 61 is an explanatory diagram illustrating a schematic configuration of a document image 211 according to the third embodiment.

  As shown in FIG. 61, the document image 211 according to the third embodiment has one or more ruled lines. Note that the document image 211 according to the third embodiment is different from the document image 11 according to the first embodiment shown in FIG. The same applies to the point.

  As shown in FIG. 61, the ruled lines of the document image 211 include two types of ruled lines: a horizontal ruled line (horizontal ruled line 241) and a vertical ruled line (vertical ruled line 242). That is, the ruled line processing unit 215 extracts (S231) or removes (S232) these two types of ruled lines.

(Regarding ruled line extraction processing (S231) and ruled line removal processing (S232))
Here, the ruled line extraction process (S231) and the ruled line removal process (S232) will be described with reference to FIG. FIG. 62 is a flowchart showing an outline of ruled line extraction processing (S231) and ruled line removal processing (S232).

  As shown in FIG. 62, the ruled line processing unit 215 first raster scans the document image 211 to detect black pixels (S241).

  Next, the ruled line processing unit 215 searches for black pixels continuously present in the horizontal direction from the position of the black pixel detected in S241, and counts the number of black pixels (S242).

  Next, the ruled line processing unit 215 determines the number of pixels counted in S242 (S243). The ruled line processing unit 215 executes integration processing (S244) if the number of pixels is equal to or greater than the predetermined value L, and deletes (changes to white pixels) the continuous black pixel area searched in S241 if the number is less than or equal to L. A deletion process (S246) is executed.

  Next, if there are continuous black pixel areas equal to the length searched in S242 near the top and bottom of the area searched in S242, the ruled line processing unit 215 regards them as the same ruled line and integrates them (S244).

  Finally, the ruled line processing unit 215 records the ruled line information integrated in S244 (for example, information on length, width, position, etc.) and deletes the searched ruled line (S246). When S246 ends, the ruled line processing unit 215 executes the black pixel detection process (S241) again.

  This is the end of the description of the process of extracting (S231) and removing (S232) the horizontal ruled line 241. In the case of the vertical ruled line 242, a change is made to search for black pixel regions that are continuous in the vertical direction in S242 shown in FIG. 62, and if there are line segments of the same length on the left and right in S244, they are integrated as the same ruled line. The ruled line processing unit 215 executes the ruled line extraction process (S231) and the ruled line removal process (S232).

  Also, the ruled line processing unit 215 according to the third embodiment has been described by taking as an example the case of extracting and removing ruled lines by executing the ruled line extraction process (S231) and the ruled line removal process (S232). For example, the ruled line extraction method can be implemented even when a general method used in document structure analysis such as OCR is used.

  Here, when the ruled line extraction processing (S231) is executed by the ruled line processing unit 215 on the document image 211 shown in FIG. 61, only the ruled line area in the document image 211 shown in FIG. Can be extracted.

  Also, FIG. 61 shows that the ruled line processing unit 215 arranges the ruled lines in the background image composed of white pixels based on the ruled line information output by executing the ruled line extraction process (S231) and the ruled line removal process (S232). Only ruled lines in the document image 211 can be reproduced. FIG. 63 is an explanatory diagram showing a schematic configuration of a document image in which only ruled lines are displayed in the document image 211 shown in FIG.

  Further, the ruled line processing unit 215 executes the ruled line extraction process (S231) and the ruled line removal process (S232), thereby removing the horizontal ruled lines 241 and the vertical ruled lines 242 from the document image 211 shown in FIG. As shown in the figure. FIG. 64 is an explanatory diagram showing an example of a schematic configuration of a document image from which ruled lines are removed from the document image shown in FIG.

  Further, the document feature data conversion unit (image processing unit) 212 executes processing such as character region extraction processing on the document image 211 from which the ruled lines shown in FIG. 64 are removed. Note that the operation of the document feature data conversion unit 211 according to the third embodiment is substantially the same as that of the document feature data conversion units 12 and 116 according to the first or second embodiment, and thus detailed description will be made. Is omitted.

(About the operation of the watermark information synthesis unit 213)
The watermark information synthesis unit (watermark information embedding unit) 213 according to the third embodiment embeds the falsification detection data Data as watermark information in the document image 211 in the same manner as the watermark information synthesis unit 13 according to the first embodiment. .

  Note that the falsification detection data Data according to the third embodiment is obtained by further adding the following data (falsification detection data Data205) to the falsification detection data Data according to the first embodiment.

  Data 205: Ruled line information (data regarding the position, width, length, etc. of the extracted ruled line area)

(Regarding the operation of the input image transformation unit 227)
Next, the operation of the third input image transformation unit 227 provided in the document input unit 220 will be described. The third input image deformation unit 227 refers to the ruled line information of the falsification detection data Data 205 with respect to the binarized input document image described in the input image deformation unit 24 of the first embodiment. Delete.

  As described in the first embodiment, the third input image transformation unit 227 creates a corrected image from the output document image 214 from an input document image from an input device (not shown) such as a scanner, and further The ruled line area included in the image area is removed from the corrected image.

  An input document image 221 acquired by an input device (not shown) is shown in FIG. The third input image transformation unit 227 creates a corrected image for the input document image 221 shown in FIG. 65, and further removes ruled lines. FIG. 65 is an explanatory diagram showing a schematic configuration of the input document image 221 acquired by the input device.

  Further, when the third input image deformation unit 227 removes the ruled line area including the horizontal ruled line 241 and the vertical ruled line 242, the result is as shown in FIG. 66 is an explanatory diagram illustrating an example of a schematic configuration of an input document image in which ruled lines are deleted from an image obtained by correcting the input document image 211 illustrated in FIG. 65 by binarization or the like.

  The processing operation of the third input image deformation unit 227 is substantially the same as the processing other than the ruled line deletion processing compared to the operation of the input image deformation unit 24 according to the first embodiment, and thus detailed description thereof is omitted. To do.

  In addition, the falsification detection process when there is a ruled line in the document image 211 according to the third embodiment has been described, but the document image 11 according to the first embodiment or the second embodiment is applied. Even when ruled lines exist in the document image 111, the present invention can be implemented. That is, the ruled line extraction process (S231), the ruled line removal process (S232) by the ruled line processing unit 215 according to the third embodiment, and the ruled line removal process by the third input image deformation unit 227 are performed in the first implementation. The second embodiment or the second embodiment can be implemented by performing substantially the same processing.

  As described above, the falsification detection process according to the third embodiment is substantially the same as the falsification detection process (S53) by the falsification detection unit (falsification determination unit) 25 according to the first embodiment. Therefore, detailed description is omitted.

  The falsification detection process according to the third embodiment has been described above. With this configuration, even when ruled lines exist in the document image, the falsification detection is performed without requiring special processing or work by the user. The apparatus can automatically determine whether the input document image has been tampered with automatically. Further, since the position information for specifying the ruled line is not embedded in the document image in the image format, the position information related to the ruled line is embedded as the watermark information with the coordinate value or the like in the document image, so that the data amount of the watermark information is extremely increased. Can be prevented.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, this invention is not limited to this example. It is obvious for those skilled in the art that various changes or modifications can be envisaged within the scope of the technical idea described in the claims, and these are naturally within the technical scope of the present invention. It is understood that it belongs.

  In the above-described embodiment, the case where the falsification detection data Data is embedded in the document image as watermark information including a background pattern or the like has been described as an example. However, the present invention is not limited to such an example. For example, the present invention can be implemented even when the falsification detection data Data is embedded in a document image as a two-dimensional barcode.

  The present invention is applicable to.

It is a block diagram which shows the schematic structure of the document output part 10 concerning 1st Embodiment. It is a block diagram which shows an example of a schematic structure of the document input part concerning 1st Embodiment. It is a flowchart which shows an example of the outline of a process of the document characteristic data conversion part concerning 1st Embodiment. It is explanatory drawing which shows an example of the schematic structure of the document image concerning this Embodiment. It is explanatory drawing which shows the outline of the extraction process of a character area. It is explanatory drawing which shows the outline of the extraction process of a character area. It is explanatory drawing which shows the outline of the extraction process of a character area. It is explanatory drawing which shows the outline of the extraction process of a character area. It is explanatory drawing which shows an example of the outline which cuts out and reduces processing (S32) about the character area of the area | region 1 and the area | region 2 which are shown in FIG. It is explanatory drawing which shows an example of a schematic structure of the synthetic | combination image of the reduction character area concerning this Embodiment. It is explanatory drawing which shows an example of a schematic structure of the synthetic | combination image of the reduction character area concerning this Embodiment. It is explanatory drawing which shows an example of a schematic structure of the table of the falsification detection data Data2 embedded in a document image. It is a flowchart which shows an example of the schematic flow of a process of a watermark information synthetic | combination part. It is explanatory drawing which shows an example of a watermark signal. It is sectional drawing which looked at the change of the pixel value of Fig.20 (1) from the direction of arctan (1/3). It is explanatory drawing which shows an example of a watermark signal, (3) shows the unit C, (4) shows the unit D, (5) shows the unit E. FIG. 23 (1) is an explanatory diagram showing a case where the unit E is defined as a background unit and arranged as a background of the original image 11 without gaps. 23 (2) shows an example in which the unit A is embedded in the background image of FIG. 23 (1), and FIG. 23 (3) shows an example in which the unit B is embedded in the background image of FIG. 23 (1). Show. It is explanatory drawing which shows an example of the symbol embedding method to an original image. 5 is a flowchart illustrating a method for embedding watermark information in an original image. It is explanatory drawing shown about the method of embedding watermark information in an original image. It is explanatory drawing which shows an example of a watermarked document image. It is explanatory drawing which expanded and showed a part of FIG. It is explanatory drawing which shows an example of a schematic structure of an input document image. It is explanatory drawing of the signal detection filtering process (step S310) in 1st Embodiment. It is explanatory drawing of the signal position search process (step S320) in 1st Embodiment. It is explanatory drawing of the signal boundary determination process (step S340) in 1st Embodiment. It is explanatory drawing which shows an example of information restoration. It is explanatory drawing which shows an example of the decompression | restoration method of a data code. It is explanatory drawing which shows an example of the decompression | restoration method of a data code. It is explanatory drawing which shows an example of the decompression | restoration method of a data code. It is explanatory drawing which shows an example of the decompression | restoration method of a data code. It is a flowchart which shows the outline of a process of a pseudo document image generation part. It is explanatory drawing which shows the outline of the production | generation process of the pseudo document image by the pseudo document image production | generation part. It is a flowchart which shows the outline | summary of operation | movement of an input image deformation | transformation part. It is explanatory drawing which displayed the signal unit position detected in 1st Embodiment on the input image. It is explanatory drawing which shows the example of a detection of an approximate straight line. It is explanatory drawing which shows the example of the result of having performed linear approximation. It is explanatory drawing which shows correction | amendment of inclination. It is explanatory drawing which shows the correction | amendment of a position. It is explanatory drawing which shows the example of the intersection of a straight line. It is explanatory drawing which shows the example of a response | compatibility of the position of an input image and a correction image. It is explanatory drawing which shows the example of the matching method of an input image and a correction image. It is explanatory drawing which shows an example of a schematic structure of the input document image at the time of further reducing and enlarging the corrected and binarized input document image. It is a flowchart which shows the outline of a process of the tampering detection part concerning 1st Embodiment. It is explanatory drawing which shows an example of a schematic structure of a difference image. It is explanatory drawing which shows an example of a schematic structure of a difference image. It is a block diagram which shows an example of a schematic structure of the document output part concerning 2nd Embodiment. It is a block diagram which shows an example of a schematic structure of the document input part concerning 2nd Embodiment. It is a flowchart which shows the outline of a process of the 2nd document feature data conversion part concerning 2nd Embodiment. It is explanatory drawing which shows an example of the schematic structure of the importance of the description item block concerning 2nd Embodiment. It is explanatory drawing which shows an example of the result of the number allocation process by the 2nd document feature data conversion part concerning 2nd Embodiment. It is explanatory drawing which shows an example of the outline of a cutting-out and reduction process (S132) of a character area. It is explanatory drawing which shows an example of the schematic structure of the synthesized image concerning 2nd Embodiment. It is explanatory drawing which shows schematic structure of the alteration detection data Data102 concerning 2nd Embodiment. It is explanatory drawing which shows an example of a schematic structure of the alteration detection data Data104 concerning 2nd Embodiment. It is explanatory drawing which shows an example of a schematic structure of a pseudo document image. It is explanatory drawing which shows the outline of operation | movement of the 2nd input image deformation | transformation part. It is a block diagram which shows schematic structure of the document output part concerning 3rd Embodiment. It is a block diagram which shows the schematic structure of the document input part concerning 3rd Embodiment. It is a flowchart which shows the outline of a process of the ruled line process part concerning 3rd Embodiment. It is explanatory drawing which shows schematic structure of the document image concerning 3rd Embodiment. It is a flowchart which shows the outline of a ruled line extraction process (S231) and a ruled line removal process (S232). FIG. 62 is an explanatory diagram illustrating an example of a schematic configuration of a ruled line region in the document image illustrated in FIG. 61. FIG. 62 is an explanatory diagram illustrating an example of a schematic configuration of a document image from which ruled lines are removed from the document image illustrated in FIG. 61. It is explanatory drawing which shows schematic structure of the input document image 221 acquired with the input device. FIG. 66 is an explanatory diagram illustrating an example of a schematic configuration of an input document image obtained by deleting ruled lines from an image obtained by correcting the input document image 211 illustrated in FIG. 65 by binarization or the like.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Document output part 11 Document image 12 Document feature data conversion part 13 Watermark information synthetic | combination part 14 Output document image 21 Input document image 22 Watermark information extraction part 23 Pseudo-document image generation part 24 Input image deformation | transformation part 25 Tampering detection part 26 Tampering detection result

Claims (11)

An image processing unit that inputs an image and executes a process of cutting out a specific region of the region for determining whether or not one or more tampering has been performed from the input image;
At least one of area information regarding the position and / or size of the specific area cut out by the image processing unit or a partial image of the image corresponding to the specific area is embedded in the image as watermark information to generate an output image A watermark information embedding unit;
With
The image processing unit executes a process of cutting out the specific area, and further reduces the partial image of the image corresponding to the cut out specific area at a reduction ratio determined according to the importance for each specific area. A watermarked image output apparatus characterized by executing a reduction process.   The watermarked image output apparatus according to claim 1, wherein the image processing unit dynamically changes the importance set for each specific area. An image processing unit that inputs an image and executes a process of cutting out a specific region of the region for determining whether or not one or more tampering has been performed from the input image;
At least one of area information regarding the position and / or size of the specific area cut out by the image processing unit or a partial image of the image corresponding to the specific area is embedded in the image as watermark information to generate an output image A watermark information embedding unit;
With
The watermark image output apparatus, wherein the image processing unit collects each of the one or more reduced images and generates one composite image. The image processing unit executes a process of cutting out the specific area, and further executes a reduction process of reducing a partial image of the image corresponding to the cut out specific area at a predetermined reduction rate;
The watermark information includes the area information and further includes a partial image of the image corresponding to the specific area or a reduced image obtained by reducing a partial image of the image corresponding to the specific area. The watermarked image output apparatus according to claim 1, further comprising a rate . 4. The watermarked image according to claim 1, wherein the specific area is a character area including at least one of a character, a symbol, or a figure. Output device. The watermarked image output apparatus further includes a straight line / broken line processing unit that extracts a straight line and / or broken line area having a predetermined length included in the image,
The watermark information embedding unit further embeds, as watermark information, straight line area information relating to the position and / or size of a straight line and / or broken line area extracted by the straight line / broken line processing unit, into the image. 4. The watermarked image output apparatus according to claim 1, 2, or 3 . At least one of area information on the position and / or size of one or more specific areas for determining whether or not the image has been tampered with, or a partial image of the image relating to the specific area is used as watermark information. A watermark information extraction unit that inputs an output image generated by being embedded in the image and extracts the watermark information embedded in the input output image;
An image transformation unit that processes the input output image and transforms the input image into an input image having substantially the same size as the image;
A pseudo image generation unit that generates a pseudo image at least similar to a partial image of the specific region existing in the image;
A falsification determination unit that generates a difference image including a difference between the input image and the pseudo image and determines that an output image that is a deformation source of the input image has been falsified when a difference area exists in the difference image When;
With
When the pseudo image generation unit generates a background image having substantially the same size as the image, and the watermark information includes a partial image of the image corresponding to the specific area, the image is directly arranged in the background image according to the area information. Or when the watermark information includes a reduced image corresponding to the specific area, the image is first enlarged at a reciprocal of the reduction ratio reduced to the reduced image, and the enlarged reduced image is arranged according to the area information; A watermarked image input apparatus, wherein the pseudo image is similar to the image by generating the pseudo image. At least one of area information on the position and / or size of one or more specific areas for determining whether or not the image has been tampered with, or a partial image of the image relating to the specific area is used as watermark information. A watermark information extraction unit that inputs an output image generated by being embedded in the image and extracts the watermark information embedded in the input output image;
An image transformation unit that processes the input output image and transforms the input image into an input image having substantially the same size as the image;
A pseudo image generation unit that generates a pseudo image at least similar to a partial image of the specific region existing in the image;
A falsification determination unit that generates a difference image including a difference between the input image and the pseudo image and determines that an output image that is a deformation source of the input image has been falsified when a difference area exists in the difference image When;
With
The image transformation unit cuts out an area that matches the specific area in the input output image based on area information regarding the position and / or size of the specific area included in the watermark information, and corresponds to the specific area The partial image of the clipped area is reduced at a reduction ratio included in the watermark information, and the partial image of the clipped area is enlarged at a magnification that is the inverse of the reduction ratio. A watermarked image input device characterized by being arranged at a position. At least one of area information on the position and / or size of one or more specific areas for determining whether or not the image has been tampered with, or a partial image of the image relating to the specific area is used as watermark information. A watermark information extraction unit that inputs an output image generated by being embedded in the image and extracts the watermark information embedded in the input output image;
An image transformation unit that processes the input output image and transforms the input image into an input image having substantially the same size as the image;
A pseudo image generation unit that generates a pseudo image at least similar to a partial image of the specific region existing in the image;
A falsification determination unit that generates a difference image including a difference between the input image and the pseudo image and determines that an output image that is a deformation source of the input image has been falsified when a difference area exists in the difference image When;
With
The image deforming unit reduces the partial image of the clipped area at a reduction ratio determined according to the importance for each specific area, and further, the partial image of the clipped area at a magnification that is a reciprocal of the reduction ratio. After the image is enlarged, the watermarked image input device is arranged at the original position in the output image. At least one of area information on the position and / or size of one or more specific areas for determining whether or not the image has been tampered with, or a partial image of the image relating to the specific area is used as watermark information. A watermark information extraction unit that inputs an output image generated by being embedded in the image and extracts the watermark information embedded in the input output image;
An image transformation unit that processes the input output image and transforms the input image into an input image having substantially the same size as the image;
A pseudo image generation unit that generates a pseudo image at least similar to a partial image of the specific region existing in the image;
A falsification determination unit that generates a difference image including a difference between the input image and the pseudo image and determines that an output image that is a deformation source of the input image has been falsified when a difference area exists in the difference image When;
With
The watermark information extraction unit included in the watermarked image input unit outputs, as watermark information, straight line / broken line area information regarding the position and / or size of a straight line and / or broken line area having a predetermined length included in the image. Extracted from the image;
The watermarked image input apparatus further includes a straight line / broken line removal unit that removes a straight line and / or broken line area present in the input output image based on the straight line / broken line area information included in the watermark information. A watermarked image input device. The watermarked area according to claim 7, 8, 9, or 10, wherein the specific area is a character area including at least one of a character, a symbol, or a figure. Image input device.

Images