JP2006261824A - Falsification detecting system and method for detecting falsification - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a falsification detecting system and a method for detecting falsification capable of properly deciding the presence of the falsification by using the feature quantity of an image before printing and the feature quantity of the image after the printing. <P>SOLUTION: The falsification detecting system which decides the presence of the falsification of the printing contents of a print, is provided with: an altered image preparation section 118 altering a read image prepared by reading the print with a printed original image and preparing a plurality of altered images obtained by altering respective read images differently; a feature quantity measuring section 102 measuring each feature quantity of the plurality of altered images; a feature quantity difference computing section 122 obtaining the differential value of the feature quantity of the original image and the feature quantity of each altering image; and a decision section 124 deciding whether or not the read image is falsified on the basis of values obtained by the feature quantity difference computing section. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a falsification detection system and a falsification detection method for determining whether or not falsification of printed content of a printed matter is present.

  There is a need for a device that can determine whether a document printed on a paper medium has been tampered with. The following technologies are known as devices that can determine whether or not tampering has occurred.

[1] JP 2000-232573 “Printing apparatus, printing method, and recording medium”
When printing the print data on at least the printed matter including the form, by printing the electronic watermark corresponding to the print data in addition to the print data, the printed file can be faithfully reproduced just by looking at the printed matter. In addition, it is possible to determine whether or not the print result has been tampered with from the electronic watermark information printed on the printed matter. The determination of tampering is performed by comparing the print result with the one printed with the digital watermark.

  According to the above-described apparatus, in order to determine whether or not tampering has occurred, it is necessary to visually compare the print content extracted from the digital watermark with the print content printed on the paper surface, and there is a possibility that the tampering may be missed due to human error. Therefore, there is a need for a method for automatically determining whether or not tampering has occurred, not by visual inspection. In order to determine automatically, a method of using the feature quantity of the original document and the feature quantity of the document after printing and determining the presence / absence of alteration based on the difference between the feature quantities can be considered. Specifically, it is possible to automatically determine the presence / absence of falsification by setting a threshold value for determining whether or not falsification has occurred and comparing the difference between the feature amounts and the threshold value.

JP 2000-232573 A

  However, there is a problem that the threshold value must be set to an appropriate value when the tampering determination is automatically performed using the feature amount. In particular, when a printed document is input to the computer again by an input device such as a scanner, many noise components are included in the input image due to image deformation such as rotation that occurs during input. Therefore, the difference between the feature value of the original document and the feature value of the input image becomes large, and depending on the set threshold value, erroneous detection of falsification that may determine that an unmodified document has been altered may occur. There is sex. On the other hand, if a value that can prevent false detection of falsification is set as a threshold, there is a possibility that a falsification miss that determines that a falsified document has not been falsified may occur.

  Therefore, the present invention has been made in view of such problems, and the object of the present invention is to appropriately determine whether or not tampering has occurred using the feature amount of the image before printing and the feature amount of the image after printing. It is an object of the present invention to provide a falsification detection system and a falsification detection method that can be performed.

  In order to solve the above problems, according to an aspect of the present invention, there is provided a falsification detection system for determining whether or not falsification of the print contents of a printed matter. The falsification detection system includes a change image creation unit that modifies a read image created by reading a printed matter on which an original image is printed, and creates a plurality of change images in which different changes are made to the read image; A feature amount measuring unit for measuring each feature amount of the plurality of changed images; a feature amount difference calculating unit for obtaining a difference value between the feature amount of the original image and the feature amount of each changed image; and a feature amount difference calculating unit A determination unit that determines whether or not the read image has been tampered with based on the obtained value.

  According to the above invention, a plurality of images whose feature values are compared with the original image are created based on the read image by the modified image creation unit. The feature quantity of each created image is measured by the feature quantity measurement unit, and the difference value between each measured feature quantity and the feature quantity of the original image is obtained by the feature quantity difference calculation unit. The determination unit determines whether or not the read image is different from the original image based on the value obtained by the feature amount difference calculation unit, that is, whether or not falsification or alteration has been performed. The determination unit performs the determination based on the result of comparing the feature amount of the original image with the feature amount of the created changed image. Specifically, for example, the minimum difference obtained by the feature amount difference calculation unit is used. The determination may be performed based on the value, or may be performed based on the sum or product of the values obtained by the feature amount difference calculation unit. In other words, the falsification detection system according to the present invention searches for a modified image having a feature amount closest to the feature amount of the original image from a plurality of modified images created based on the read image, and the modified image and the original image are searched for. Whether or not the image is the same, or is not exactly the same but has a difference that is not recognized as being altered or altered, is determined based on the difference in the feature amount. It is determined whether or not has been tampered with. According to such a configuration, the original image on the scanned image is rotated or distorted due to the state of ink or toner when printing the original image, the folds or wrinkles of the printed material, or the deviation of the direction when the printed material is read by the scanner. Even when an unintended change such as ink or toner bleeding or blurring is made, it is possible to prevent erroneous detection of tampering.

  The feature amount measuring unit may measure a value output as a feature amount by applying one or more feature extraction filters that strongly react to a specific shape to each image.

  The determination unit may determine whether the read image has been falsified based on a minimum value among the plurality of difference values obtained by the feature amount difference calculation unit and a predetermined falsification threshold. According to this configuration, the presence or absence of falsification of the read image is determined based on the changed image having the feature amount closest to the feature amount of the original image created based on the read image. Therefore, even if an inevitable unintended change is made when printing the original image or reading the printed matter, it is determined that the read image that has not been tampered with has been tampered with due to the change. Can be prevented.

  The changed image creating unit may change the pixel density value of the read image. According to such a configuration, when an original image is printed or when a printed material is read, if an unintended change that is not falsified is made to the density value of some or all of the pixels in the image, the changed image creation unit Thus, by changing the density value of the pixel of the read image, an unintended change made during printing or the like can be removed.

  The modified image creation unit may binarize the read image based on a predetermined binarization threshold.

  The changed image creation unit may be capable of changing the binarization threshold. According to this configuration, the changed image creation unit can create a plurality of binary images from the read image by changing the threshold value.

  The modified image creation unit may replace a white pixel adjacent to a black pixel with a black pixel.

  The modified image creation unit may replace black pixels adjacent to white pixels with white pixels.

  The modified image creation unit may rotate, enlarge, reduce, and / or translate the read image. According to such a configuration, when the original image is printed or when the printed material is read, as an unintentional change that has not been tampered with, the image is tilted, distorted, misaligned, or the image is reduced or enlarged. Unintentional changes made during printing can be removed by affine transformation of the read image by the modified image creation unit.

  In the read image, position information of an arbitrary region of the original image and a feature amount of the arbitrary region are embedded in association with each other, and the falsification detection system performs the position information embedded in the read image and the feature. An extraction unit that extracts an amount; and a position information acquisition unit that acquires position information of a region corresponding to an arbitrary region of the original image in the read image based on the position information extracted by the extraction unit; Also good. At this time, the modified image creation unit creates a modified image from an area on the read image corresponding to an arbitrary area of the original image based on the position information acquired by the position information acquisition unit, and calculates the feature amount difference. The unit may obtain a difference value between the feature amount extracted from the read image by the extraction unit and the feature amount of each changed image. According to such a configuration, it is possible to determine whether an image in an arbitrary region of the original image has been tampered with in the read image.

  The feature amount measurement unit measures a feature amount of an arbitrary region of the original image, and the falsification detection system embeds the positional information of the arbitrary region of the original image and the feature amount of the arbitrary region in an embedded manner in the original image. A feature amount embedding unit; a coordinate information embedding unit that embeds coordinate information in an original image in association with a point specified by coordinates included in position information; an image output unit that outputs an original image in which coordinate information is embedded; A reading unit that reads the printed material on which the output original image is printed and creates a read image; and a coordinate information extraction unit that extracts embedded coordinate information from the read image; The position information acquisition unit is configured to select an arbitrary image of the original image in the read image based on the position information extracted by the extraction unit and the coordinate information extracted by the coordinate information extraction unit. It may acquire the position information of an area corresponding to the frequency. According to such a configuration, even when a margin is created when the original image is printed, position information of an area on the read image corresponding to an arbitrary area of the original image can be acquired by coordinate information inserted into the original image. it can.

  In order to solve the above problems, according to another aspect of the present invention, there is provided a falsification detection method for determining whether or not falsification of the print contents of a printed matter. The falsification detection method includes a modified image creation step of creating a plurality of modified images in which different modifications are made to the scanned image by modifying the scanned image created by scanning the printed material on which the original image is printed; A feature amount measuring step for measuring each feature amount of the plurality of changed images; a feature amount difference calculating step for obtaining a difference value between the feature amount of the original image and the feature amount of each changed image; and a feature amount difference calculating unit A determination step of determining whether the read image has been tampered with based on the obtained value.

  As described above, according to the present invention, there is provided a falsification detection system and a falsification detection method capable of appropriately performing falsification presence / absence determination using a feature amount of an image before printing and a feature amount of an image after printing. It can be done.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the present specification and drawings, components having substantially the same functional configuration are denoted by the same reference numerals, and redundant description is omitted.

  The falsification detection system according to the present embodiment is a system that detects whether or not an image after printing has been falsified using the feature amount of the image before printing and the feature amount of the image after printing. This system measures the feature amount of an image before printing (hereinafter also referred to as an original image) and the feature amount of an image created by reading a printed image (hereinafter also referred to as a read image). A function to measure, a function to compare the feature quantity of the original image and the feature quantity of the read image, and a function to determine whether or not the read image has been falsified as a result of the comparison are provided. In the falsification detection system according to the present embodiment, each computer is provided with the above-described functions for realizing a falsification detection function. However, a plurality of functions are not provided in one computer as in this embodiment, but each function may be distributed to two or more computers.

  FIG. 1 is a block diagram showing a functional configuration of the falsification detection system 10. The falsification detection system 10 includes, for example, a feature amount measuring unit 102, a feature amount embedding unit 104, a coordinate information embedding unit 106, an image output unit 108, a reading unit 110, an extracting unit 112, and a coordinate information extracting unit 114. A position information acquisition unit 116, a modified image creation unit 118, a feature amount difference calculation unit 122, a determination unit 124, and the like. As described above, each function may be provided in one computer, or may be distributed among a plurality of computers. For example, a portion that performs processing on the original image (feature amount measurement unit 102, feature amount embedding unit 104, coordinate information embedding unit 106, and image output unit 108) and a portion that performs processing on the read image (reading unit 110, extraction unit 112). , The coordinate information extraction unit 114, the position information acquisition unit 116, the changed image creation unit 118, the feature amount measurement unit 102, the feature amount difference calculation unit 122, and the single bottom portion 124) may be provided in different computers.

  The feature amount measuring unit 102 has a function of measuring the feature amount of the input original image 1002. The original image 1002 is an image that is a source of a read image to be subjected to tampering detection. As a method for measuring the feature amount, a method of obtaining a feature amount of each image by performing a filtering process with a Gabor filter or a known filter on each image and obtaining an average value of the filter output may be applied. it can. Further, the feature amount may be measured in more detail by dividing the image into a plurality of regions and measuring the feature amount for each of the divided regions. Besides the feature quantity measurement by filtering, a method of performing pattern matching by generating a reduced image and measuring the degree of coincidence can be applied. The feature quantity measured from the image is almost the same feature quantity in the same image. However, even if the image is the same, the image before printing and the scanned image of the printed matter may be contaminated, smeared, deformed, etc. depending on the performance of the printer or scanner. In some cases, the feature amount of the image and the feature amount of the read image are different. In the present embodiment, the feature amount measuring unit 102 designates an arbitrary region from the original image 1002 and measures the feature amount of the designated region. The feature amount measuring unit 102 may specify the entire area of the original image 1002 or a part of the original image 1002 as an arbitrary area.

  The feature amount measured by the feature amount measuring unit 102 and the position information specifying the region where the feature amount is measured are provided to the feature amount embedding unit 104. The feature amount embedding unit 104 associates the provided feature amount with the position information and embeds the feature amount in the original image 1002 as a digital watermark.

  The position information for specifying the region where the feature amount is measured by the feature amount measuring unit 102 is also provided to the coordinate information embedding unit 106. The coordinate information embedding unit 106 embeds the coordinate information in the original image 1002 as a digital watermark in association with the point specified by the coordinates included in the provided position information. Specifically, as shown in FIG. 3, information on coordinates included in the position information acquired from the feature quantity measurement unit 102 is embedded in a point on the original image specified by the coordinates. According to the illustrated example, the coordinates (200, 200) of the start point 2002 are embedded as a digital watermark at the start point 2002 designated by the feature amount measuring unit 102 and indicating the start position of the region where the feature amount is measured. . Similarly, the coordinates (300, 300) of the end point 2004 are embedded as an electronic watermark at the end point 2004 indicating the end position of the region where the feature amount is measured.

  Here, the digital watermark will be described. For example, as disclosed in Japanese Patent Application Laid-Open No. 2003-209676, “Digital Watermark Embedding Device, Digital Watermark Detection Device, Digital Watermark Embedding Method, and Digital Watermark Detection Method”, the digital watermark technology uses an arbitrary dot pattern. Is a technology that embeds the image in paper and transmits information as a digital watermark in addition to physical information for visual inspection.

  Digital watermarking technology not only embeds arbitrary information on paper, but also the small unique patterns to be embedded themselves indicate the positional relationship between the patterns or relative to the paper, and after the image is read by a scanner or the like. This is used as a reference for correcting the image.

  FIG. 11 is an explanatory diagram for explaining an arbitrary pattern embedded in a digital watermark. Referring to FIG. 11, an image based on such digital watermark is represented as a set of N unit patterns composed of a plurality of small dots. Each unit pattern corresponds to a numerical value of 0 to N−1. For example, the unit pattern 902 represents a numerical value 0 and the unit pattern 904 represents a numerical value 1. The unit pattern can be expressed in black and white or color.

  FIG. 12 is an external view for explaining an image (paper) printed with a digital watermark. Referring to the entire sheet of FIG. 12, the image based on the digital watermark can be divided into a main area 910, an attribute information area 912, and a boundary information area 914 according to the role of the image. The main area 910 is a main area for embedding information. The attribute information area 912 stores information necessary for reading the information embedded in the main area 910, such as the size of the area in which the digital watermark is embedded and the information encoding method in the main area 910. A predetermined specific unit pattern is printed in the boundary information area 914, and the reading side can identify the boundary of the digital watermark by this area.

  On the reading side such as a scanner, the paper printed in this way (hereinafter simply referred to as digital watermark paper) is scanned to detect individual unit patterns embedded in the paper. This detection is performed using template matching or a two-dimensional filter (for example, a two-dimensional Gabor filter) having a filter characteristic for an arbitrary directionality and shape.

  Further, when embedding a digital watermark in the paper, specific unit patterns are regularly arranged at specific positions in the boundary information area 914 and the main area 910. Thus, on the paper reading side, these unit patterns for position detection can be extracted and correction can be performed based on the position information. Therefore, the original image can be restored without performing troublesome manual correction even if the scanned data is displaced due to displacement or rotation or biased due to wrinkles.

  In addition, when printing or scanning, a part of the unit pattern may be stained with, for example, ink and cannot be detected. Therefore, in the main area 910, information to be embedded is distributed at a plurality of locations, or is embedded with an error resilience code (parity check code, Hamming code, BCH code, etc.) as disclosed in Japanese Patent Laid-Open No. 2003-209676. For example, even if a part of the unit pattern is missing, the information can be restored.

  Returning to FIG. 1, the description of the functional configuration of the falsification detection system 10 will be continued. The image output unit 108 prints the original image 1002 in which the feature amount, the position information of the region where the feature amount is measured, and the coordinate information of the region are embedded as a digital watermark. The image output unit 108 does not print the original image, but may output it as an electronic data to an external recording medium such as a flexible disk or a CD (Compact Disk), or connected via a communication network. It may be sent to another computer. Then, the original image 1002 that has been output to the external recording medium or transmitted to another computer may be printed by another computer or the like.

  The reading unit 110 reads a printed material 1004 on which an original image is printed, and creates a read image. Specifically, the printed matter is read through a reading unit such as a scanner, and a read image is created based on the read data. Note that the printed material on which the original image is printed may include a blank portion in addition to the original image portion, but the read image created by the reading unit 110 also includes such a blank portion. ing. In addition, the original image may be reduced during reading. As described above, the reading unit 110 can read a printed matter with a scanner, and can also photograph a printed matter with a digital camera, for example, and take the data of the photographed image to create a read image.

  The extraction unit 112 extracts the embedded information, that is, the feature amount and the position information where the feature amount is measured, from the read image created by the reading unit 110. The extraction unit 112 stores the extracted information in a storage area in the falsification detection system 10.

  The coordinate information extraction unit 114 extracts embedded coordinate information from the read image. The position information acquisition unit 116, based on the position information of an arbitrary area specified by the feature amount measurement unit 102 and the coordinate information extracted by the coordinate information extraction unit 114, an arbitrary area of the original image in the read image. The position information of the area corresponding to (hereinafter also referred to as the corresponding area) is acquired.

  Specifically, as shown in FIG. 4, the coordinate information extraction unit 114 searches embedded information that is not visible from the upper left end to the lower right end of the read image and extracts the information. According to the illustrated example, information indicating (200, 200) embedded in the start point 2002 of the read image and information indicating (300, 300) embedded in the end point 2004 are extracted. The position information acquisition unit 116 acquires the coordinates in the read image of the point from which the information is extracted by the coordinate information extraction unit 114, is extracted by the extraction unit 112, and is based on the information stored in the storage area in the falsification detection system 10. Thus, read image area information is created. According to the illustrated example, the start point 2002 from which the information (200, 200) is extracted is a point specified by coordinates (368, 248) on the read image, and the information (300, 300) is The extracted end point 2004 is a point specified by coordinates (435, 315) on the read image. Therefore, the area specified by the coordinates (200, 200) and (300, 300) in the original image corresponds to the area specified by the coordinates (368, 248) and (435, 315) on the read image. I understand that. Therefore, the position information acquisition unit 116 detects the alteration by associating the position information “368, 248, 435, 315” specifying the corresponding region with the feature amount of the region extracted by the extraction unit 112. The data is stored in a storage area in the system 10.

  As described above, the coordinate information embedding unit 106, the coordinate information extraction unit 114, and the position information acquisition unit 116 use the coordinate information to obtain a corresponding area between the original image and the read image, so that the original image is not included. The area on the read image that accurately corresponds to any area of the original image, even if the margin is formed in the read image, or if the read image is misaligned due to a shift in the reading direction when reading the printed matter Can be specified.

  The modified image creation unit 118 creates a plurality of modified images in which different modifications are made to the scanned image by modifying the scanned image. Specifically, the changed image creation unit 118 calculates the pixel density value for an area on the read image corresponding to an arbitrary area of the original image from which the feature amount is measured, which is obtained by the position information acquisition unit 116. Or change affine transformation to create multiple modified images. Details of creation of the changed image will be described later. Note that the changed image creation unit 118 may perform density value change and affine transformation on the entire read image.

  A feature amount of each of the created change images is measured by the feature amount measuring unit 102 described above.

  The feature amount difference calculating unit 122 calculates the feature amount of an arbitrary region of the original image measured by the feature amount measuring unit 102 and each of the feature amounts of the plurality of changed images measured by the feature amount measuring unit 102 in the same manner. It has a function for obtaining a difference value. Specifically, the feature amount difference calculation unit 122 obtains a difference value between each of the plurality of feature amounts, which are measurement results by the feature amount measurement unit 102, and the feature amount extracted from the read image by the extraction unit 112.

  The determination unit 124 determines the presence / absence of falsification based on the minimum value among the plurality of difference values calculated by the feature amount difference calculation unit 122. Specifically, the minimum difference value is compared with a threshold value for falsification detection (hereinafter referred to as a falsification detection threshold value). If the difference value is smaller than the falsification detection threshold value, the image in that area is falsified. If the difference value is equal to or greater than the falsification detection threshold, it is determined that the falsification has occurred.

  The functional configuration of the falsification detection system 10 has been described above. Next, the flow of processing of the falsification detection system 10 will be described with reference to FIG.

  First, an image is input to the falsification detection system 10 as an original image (S102). In the present embodiment, an image in which a plurality of characters are drawn side by side is used as the original image, and the falsification detection system 10 determines whether the characters drawn in the original image have been falsified in the read image. Determine whether or not. When the original image is input, the feature quantity measurement unit 102 of the falsification detection system 10 measures the feature quantity of the original image (S104). In the present embodiment, the feature amount of a region including any one character among a plurality of characters drawn in the original image is measured. The measured feature amount, the position information of the region where the feature amount is measured, and the coordinate information included in the position information are embedded in the original image by the feature amount embedding unit 104 and the coordinate information embedding unit 106 (S106). The original image is printed out by the image output unit 108 after the information is embedded in S106 (S108).

  Next, the printed matter created in S108 on which the original image is printed is read by the reading unit 110 to create a read image (S110). The embedded information is extracted from the created read image by the extraction unit 112 and the coordinate information extraction unit 114, and the read image corresponding to the region on the original image in which the feature amount is measured by the position information acquisition unit 116. The upper area is specified (S112).

  When the corresponding area is specified in S112, a modified image is created by the modified image creating unit 118 (S114), and the feature amount of the created modified image is measured by the feature amount measuring unit 102 (S116). The feature value difference calculation unit 122 obtains a difference value between the feature value of the original image extracted by the extraction unit 112 in S112 and the feature value of the changed image measured in S116 (S118). Steps S114 to S118 are executed a plurality of times, and difference values between the created feature amounts of the plurality of changed images and the feature amounts of the original image are respectively obtained. Then, based on the minimum value among the plurality of difference values obtained in S118 executed a plurality of times, the determination unit 124 determines whether or not tampering has occurred (S120).

  The processing flow of the falsification detection system 10 has been described above. Next, the modified image creation processing by the modified image creating unit 118 will be described in detail with reference to FIGS.

  The image from which the modified image creating unit 118 creates the modified image is an image of a region corresponding to an arbitrary region (region in which the feature amount is measured) of the original image in the read image. As described above, the coordinate information embedding unit 106, the coordinate information extraction unit 114, and the position information acquisition unit 116 use the coordinate information to obtain an area corresponding to an arbitrary area of the original image in the read image. For this reason, it is possible to specify a region on the read image that almost exactly corresponds to an arbitrary region of the original image. If the corresponding area can be identified almost accurately, the feature value of an arbitrary area of the original image and the feature value of the area on the read image corresponding to the area are almost equal unless the area of the read image is falsified. The difference value between the two feature amounts should be smaller than the falsification threshold.

  However, when an original image is printed by the image output unit 108, an unintended change is added to an arbitrary area of the original image, or when a printed matter is read by the reading unit 110, an unintended change is added to a corresponding area on the read image. May be. Specifically, for example, when the original image is printed, the ink bleeds and the line width of characters, figures, symbols, etc. drawn on the original image becomes wide, or the characters are faded due to lack of ink or toner, The width of the line may be narrowed. In addition, noise may occur in the scanned image due to dirt on the printed material or dust attached to the printed material when the printed material is read. In addition, when printing is performed with an electronic watermark embedded in the original image, there may be a large difference in the feature quantity between the original image and the read image due to the digital watermark. Can be. In addition, when a printed material is read, there are cases where characters, figures, symbols, etc. on the scanned image are tilted or distorted due to a deviation in the reading direction or wrinkles of the printed material. In this way, if an unintended change is made to the original image or the read image, there may be a large difference in the feature quantity between the original image and the read image even if the read image has not been tampered with. is there. As a result of the difference between the feature value of the original image and the feature value of the read image, the difference value between the feature values becomes larger than the falsification detection threshold, and the read image is not actually falsified. Regardless, the determination unit 124 may determine that there has been tampering. Therefore, the modified image creation unit 118 attempts to create a modified image that is close to the state before the unintended modification is made by adding a plurality of patterns to the read image. If the determination unit 124 determines whether or not falsification has occurred based on the feature amount of the changed image closest to the state before the unintended change is added and the feature amount of the original image, the tampering is not actually performed. Nevertheless, it can be prevented that the tampering is determined. Hereinafter, specific description will be given with reference to the drawings.

  FIG. 5 shows an example in which the width of the character line in the read image is wider or narrower than that of the original image in the read image as an unintended change. As illustrated, the feature amount of the original image 202 is 6.0. On the other hand, the feature amount of the read image 204 with the expanded line width is 4.7, and the feature amount of the read image 206 with the contracted line width is 7.8. Differences from the feature amount of the original image are 1.3 and 1.8, respectively, and when the alteration detection threshold is 1.0, both of the read images 204 and 206 are determined to be altered.

  Therefore, a plurality of pattern changes are added to the read images 204 and 206 by the modified image creation unit 118, and a plurality of modified images are created. FIG. 6A shows a case where a modified image is created based on a read image 204 with an expanded line width. As a result of the expansion processing by the modified image creating unit 118 so that the width of the character line in the scanned image 204 is widened, a modified image 204a of the scanned image 204 is created. In addition, as a result of the contraction process performed by the modified image creation unit 118 so that the width of the character line in the scanned image 204 becomes narrower, another modified image 204b of the scanned image 204 is created. The expansion process and the contraction process will be described later with reference to FIGS. The feature amount of the changed image 204a is 4.0, and the difference from the feature amount of the original image is 2.0. On the other hand, the feature amount of the changed image 204b is 5.9, and the difference from the feature amount of the original image is 0.9. According to the example shown in the figure, only two modified images 204a and 204b have been created. Therefore, the determination unit 124 determines whether or not tampering has occurred based on 0.1 as the first value of the two difference values. I do. As a result, in this example, since the falsification detection threshold is 1.0, it is determined that there is no falsification. The modified image creation unit 118 may create a plurality of modified images by changing the degree of expansion or contraction. In addition, the difference value between the feature value of the read image 204 before the change and the feature value of the original image may be included in the difference value of the plurality of feature values for obtaining the minimum value.

  FIG. 6B shows a case where a changed image is created based on the read image 206 whose line width is contracted. As a result of the expansion processing by the modified image creating unit 118 so that the width of the character line in the scanned image 206 is widened, a modified image 206a of the scanned image 206 is created. In addition, as a result of the contraction process performed by the modified image creating unit 118 so that the width of the character line in the scanned image 206 is narrowed, another modified image 206b of the scanned image 206 is created. The feature amount of the changed image 206a is 5.8, and the difference from the feature amount of the original image is 0.2. On the other hand, the feature amount of the changed image 206b is 8.6, and the difference from the feature amount of the original image is 2.6. According to the example shown in the figure, only two modified images 206a and 206b have been created, so that the determination unit 124 determines whether or not tampering has occurred based on the initial value 0.2 of the two difference values. I do. As a result, in this example, since the falsification detection threshold is 1.0, it is determined that there is no falsification. The modified image creation unit 118 may create a plurality of modified images by changing the degree of expansion or contraction. Moreover, the difference value between the feature value of the read image 206 before the change and the feature value of the original image may be included in the difference value of the plurality of feature values for obtaining the minimum value.

  Next, the expansion process and the contraction process will be described with reference to FIGS. The expansion process is a process for widening the line width of characters, figures, symbols, etc. included in the image. The expansion process can be realized by changing the density value of some pixels in the image. For example, when the image to be processed is a binary image and characters, figures, symbols, etc. are represented by black pixels, as shown in FIG. A method of converting adjacent white pixels into black pixels can be used. By converting white pixels adjacent to black pixels to black pixels, the width of a line of characters or the like can be increased. In addition, the degree of expansion can be adjusted by changing the range of white pixels to be converted into black pixels. For example, according to the example of FIG. 7, as a result of converting white pixels adjacent to black pixels into black pixels in the read image 210, a modified image 210a is created. In this example, only the white pixel adjacent to the black pixel is converted, but if the white pixel adjacent to the black pixel and the white pixel adjacent to the white pixel are further converted to black pixels, It is possible to create a modified image having more black pixels than the modified image 210a, that is, a state where the line width of the character is further expanded.

  On the other hand, the shrinking process is a process for narrowing the width of lines such as characters, figures, symbols, and the like included in an image. Shrinkage processing can also be realized by changing the density values of some pixels in the image. For example, when the image to be processed is a binary image and characters, figures, symbols, etc. are represented by black pixels, as shown in FIG. A method of converting black pixels to white pixels can be used. By converting a black pixel adjacent to a white pixel into a white pixel, the width of a line such as a character can be reduced. In addition, the degree of contraction can be adjusted by changing the range of black pixels to be converted into white pixels. For example, according to the example of FIG. 7, as a result of converting a black pixel adjacent to a white pixel into a white pixel in the read image 212, a modified image 212b is created. Note that the degree of contraction can be adjusted by changing the range of black pixels to be converted into white pixels.

  In the above description, the case where the image to be processed is a binary image has been described. If the image to be processed is not a binary image, the image is first binarized, and after the image to be processed is converted to a binary image, black pixels are converted into white pixels or white pixels are converted into black pixels as described above. Can be used. Note that smoothing processing as shown in FIG. 9 may be performed before binarization of the image to be processed. The smoothing process can be performed by a median filter or a moving average method.

  As described above, as an example in which an unintended change is made to the read image, a case has been described in which the width of the character line in the read image is wider or narrower than that of the original image. Next, based on FIG. 9, a case where noise occurs in the read image as an unintended change will be described.

  The read image 204 is a part of a read image created by reading a printed matter in which position information of an arbitrary area of the original image, a feature amount of the arbitrary area, and coordinate information included in the position information are embedded as an electronic watermark. It is. Therefore, in addition to characters, figures, symbols, and the like (character “A” in the illustrated example), invisible information is embedded as a digital watermark. Due to this digital watermark, a difference occurs in the feature amount between the original image and the read image. Therefore, it is desirable to separate characters, graphics, symbols, and the like in the image from the digital watermark, and measure the feature amount with the digital watermark removed. Therefore, the modified image creation unit 118 binarizes the scanned image 204 and creates a binary image of the scanned image as the modified image. By performing binarization, it is possible to separate characters, graphics, symbols, etc. in an image from digital watermarks. The binarization of the image can be executed by converting pixels having a density value higher than a set threshold value to white pixels and converting pixels having a density value equal to or lower than the threshold value to black pixels. In this case, the lower the threshold, the higher the probability of conversion to white pixels, and the higher the threshold, the higher the probability of conversion to black pixels. For this reason, if the threshold is too high, part of the digital watermark is converted to black pixels, and there is a possibility that the character and the digital watermark are not properly separated. If the threshold value is set high, the portion of the scanned image where dirt or dust is attached to the printed material at the time of scanning is converted to black pixels. In addition, when density unevenness occurs during printing or when brightness unevenness occurs during reading, more parts are converted to black pixels. According to the illustrated example, the read image 204f binarized by setting the threshold value to 190 has a higher proportion of black pixels than 204c, 204d, and 204e binarized by a threshold value smaller than 190. Yes. On the other hand, if the threshold is too low, some characters, graphics, symbols, etc. in the scanned image are converted to white pixels, and the characters are faded or disappeared. Therefore, the changed image creation unit 118 creates a plurality of binary images while changing the binarization threshold. By changing the threshold value for binarization, it is possible to create a modified image with less noise in which the electronic watermark is properly separated from the character while preventing the blurring and disappearance of the character in the scanned image. .

  According to the illustrated example, the changed image 204 c binarized with the threshold 80, the changed image 204 d binarized with the threshold 110, the changed image 204 e binarized with the threshold 150, and the binarized with the threshold 190. In the order of the changed image 204f, the number of black pixels in the image increases. When the modified image creation unit 118 creates the modified images 204c to 204f, the feature amount measurement unit 102 measures the feature amount of each modified image. According to this example, the feature amount of the modified image 204c is 7.1, the feature amount of the modified image 204d is 6.1, the feature amount of the modified image 204e is 5.6, and the feature amount of the modified image 204f is 3.5. is there. The feature value difference calculation unit 122 obtains a difference value between the feature value 6.0 of the original image and the feature value of each modified image, and the difference value is 1.1 for the modified image 204c and the modified image 204d as illustrated. Is 0.1, the modified image 204e is 0.4, and the modified image 204f is 2.5. The determination unit determines whether or not tampering has occurred based on the minimum value among the plurality of difference values obtained by the feature amount difference calculation unit 122. According to this example, since the minimum value is 0.1 of the changed image 204d, the determination unit 124 compares 0.1 with the alteration detection threshold value 1.0. As a result of the comparison, since the difference value is smaller than the falsification detection threshold, the determination unit 124 determines that there is no falsification.

  As shown in FIG. 9, smoothing processing may be performed before binarization of the read image. The smoothing process can be performed by a median filter or a moving average method.

  As described above, by generating a plurality of binary images from the read image while changing the binarization threshold, and determining whether the read image has been tampered with based on the feature amount of the binary image, the image output unit In accordance with the density of ink and toner when the original 108 prints the original image, and the brightness of the light source of the scanner when the reading unit 110 reads the printed matter, the fluctuation of the average density value of the entire read image, printing unevenness, and reading Even if local density value fluctuations occur due to uneven brightness, it is possible to determine the presence or absence of falsification based on the read image that is closest to the original image as a result of binarization with an appropriate threshold. . In addition, the influence of the digital watermark on the feature amount of the read image can be removed. Further, even if dirt or dust related to the fluctuation of the density value of the pixel of the read image adheres to the printed matter at the time of reading, the influence can be removed by binarizing with an appropriate threshold value. Therefore, it is possible to prevent a read image that has not been actually tampered from being erroneously determined to have been tampered with due to the change in the average density value and the local density value as described above.

  As described above, the case where noise occurs in the read image as the case where an unintended change is added to the read image has been described based on FIG. Next, a case where the read image is tilted will be described with reference to FIG. 10 as an example when an unintended change is made to the read image.

  FIG. 10 shows a state in which the read image is tilted by reading the printed matter in a tilted state. When the reading unit 110 reads a printed material with a scanner, the read image may be inclined due to, for example, a deviation in the reading direction or wrinkles of the printed material. In addition, when the reading unit 110 captures a printed matter with a digital camera and captures the captured image, the read image is inclined due to, for example, a shift in the capturing direction. The example of FIG. 10 illustrates a state in which the read image is tilted due to a shift in the shooting direction when the printed material 1004 is shot with a digital camera. As a result of the tilt of the read image, the character “A” in the read image 204 is distorted. The read image in which the distortion has occurred has a feature amount of 4.2, and a difference from the original image feature amount of 6.0 is 1.8. Since the difference value 1.8 is larger than the tampering detection threshold 1.0, the read image 204 in which the distortion has occurred is determined by the determination unit 124 to be falsified.

  Therefore, the modified image creation unit 118 creates a plurality of modified images by affine transformation of the read image. According to the example shown in the figure, a modified image 204j when rotated twice in the near direction around the central horizontal axis of the read image and a modified image 204k when rotated twice in the far direction are created. The feature amount of the changed image 204j is 5.7, and the feature amount of the changed image 204k is 4.7. If the presence / absence of alteration is determined based on the changed image 204j, the determination unit 124 determines that no alteration has occurred. As the affine transformation performed by the modified image creation unit 118, it is conceivable to convert the angle, scale, and arrangement of the read image. As described above, even if the read image is tilted or distorted by creating a plurality of affine-transformed changed images from the read image by the changed image creating unit 118, the tilt and distortion are affine transformed. The presence / absence of falsification can be determined on the basis of the changed image removed in the above. For this reason, it is possible to prevent a read image that has not been tampered with from being erroneously determined to have been tampered with due to inclination or distortion caused during reading or the like. In addition to the inclination and distortion described above, changes that occur in the read image include misalignment and changes in the size of the image. The misalignment occurs due to, for example, distortion of the paper surface of the printed material at the time of reading. The change in the size of the image occurs, for example, when the scanner performs enlargement or reduction when reading a printed matter. Further, when a printed matter is photographed with a camera, the size of the photographed image differs depending on the distance from the printed matter to the camera and the degree of zooming, so the size of the read image may differ from the original image. As described above, even when the read image is misplaced or changed in size, the tamper detection system according to the present embodiment allows the read image to be affine-transformed and rotated, enlarged, reduced, or translated to obtain a plurality of images. By creating a modified image of each image, obtaining the feature value of each modified image, and determining whether or not tampering has occurred based on the difference between the feature value closest to the original image and the feature value of the original image. Can be determined. Note that rotation, enlargement, reduction, and parallel movement of the read image by affine transformation may be performed individually or in combination.

  The modified image creation processing by the modified image creating unit 118 has been described in detail above. The modified image creation unit 118 can create a plurality of modified images from one read image by performing the above-described modification image creation processing individually or in combination.

  As described above, according to the falsification detection system 10 according to the present embodiment, first, the position information of the area where the feature value is measured in the original image is embedded in association with the coordinates, so that the corresponding area in the read image can be accurately determined. Can be specified. Then, by creating multiple modified images based on the image of the accurately specified area, the intention that may have been added to the specified area when printing the original image or reading the printed matter, etc. Unnecessary changes can be removed and an image closest to the original image can be created. The feature value of each created modified image is measured, and the read image is falsified based on the feature value that has the smallest difference from the feature value of the original image, that is, based on the modified image that is closest to the original image. By determining whether or not the image has been changed, it is possible to prevent a determination error in which a read image that has not been actually falsified is determined to have been falsified due to the unintended change.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to the example which concerns. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are of course within the technical scope of the present invention. Understood.

  The present invention can be applied to a falsification detection system that determines whether or not falsification of the printed content of a printed matter is present.

It is a block diagram which shows the tampering detection system concerning embodiment of this invention. It is a flowchart which shows the tampering detection process in the embodiment. It is explanatory drawing which shows the coordinate information embedding process in the embodiment. It is explanatory drawing which shows the coordinate information extraction process in the embodiment. It is explanatory drawing which shows the example of determination of the presence or absence of falsification in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the change image creation process in the embodiment. It is explanatory drawing which shows the example of embedding of the digital watermark in the embodiment. It is explanatory drawing which shows the example of embedding of the digital watermark in the embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 102 Feature-value measurement part 104 Feature-value embedding part 106 Coordinate information embedding part 108 Image output part 110 Reading part 112 Extraction part 114 Coordinate information extraction part 116 Position information acquisition part 118 Changed image creation part 122 Feature-value difference calculation part 124 Determination part

Claims (11)

A falsification detection system for determining whether or not a print content of a printed matter has been falsified:
A modified image creating unit that modifies a scanned image created by scanning a printed matter on which an original image is printed, and creates a plurality of modified images in which different modifications are made to the scanned image;
A feature quantity measuring unit that measures the feature quantities of each of the plurality of changed images;
A feature amount difference calculation unit for obtaining a difference value between the feature amount of the original image and the feature amount of each of the changed images;
A determination unit that determines whether or not the read image has been falsified based on a value obtained by the feature amount difference calculation unit;
An alteration detection system characterized by comprising: The determination unit determines whether the read image has been falsified based on a minimum value among a plurality of difference values obtained by the feature amount difference calculation unit and a predetermined falsification threshold. The falsification detection system according to claim 1, wherein The falsification detection system according to claim 1 or 2, wherein the changed image creating unit changes a density value of a pixel of the read image. The falsification detection system according to claim 3, wherein the modified image creation unit binarizes the read image based on a predetermined binarization threshold. The falsification detection system according to claim 4, wherein the change image creation unit is capable of changing the binarization threshold. The falsification detection system according to any one of claims 3 to 5, wherein the modified image creation unit replaces white pixels adjacent to black pixels with black pixels. The falsification detection system according to any one of claims 3 to 5, wherein the modified image creation unit replaces a black pixel adjacent to a white pixel with a white pixel. The falsification detection system according to any one of claims 1 to 7, wherein the modified image creation unit rotates, enlarges, reduces, and / or translates the read image. In the read image, position information of an arbitrary area of the original image and a feature amount of the arbitrary area are embedded in association with each other,
An extraction unit that extracts the embedded position information and the feature amount from the read image;
A position information acquisition unit that acquires position information of an area corresponding to an arbitrary area of the original image in the read image based on the position information extracted by the extraction unit;
The modified image creating unit creates the modified image from an area on the read image corresponding to an arbitrary area of the original image based on the position information acquired by the position information acquiring unit,
The feature quantity difference calculation unit obtains a difference value between the feature quantity extracted from the read image by the extraction unit and the feature quantity of each of the changed images. The falsification detection system according to item 1. The feature quantity measuring unit measures a feature quantity of an arbitrary region of the original image;
A feature amount embedding unit that embeds the positional information of an arbitrary region of the original image and the feature amount of the arbitrary region in association with each other;
A coordinate information embedding unit that embeds the coordinate information in the original image in association with a point specified by the coordinates included in the position information;
An image output unit for outputting an original image in which the coordinate information is embedded;
A reading unit that reads a printed matter on which the output original image is printed, and creates the read image;
A coordinate information extraction unit that extracts the embedded coordinate information from the read image;
The position information acquisition unit, based on the position information extracted by the extraction unit and the coordinate information extracted by the coordinate information extraction unit, a position of an area corresponding to an arbitrary area of the original image in the read image The falsification detection system according to claim 9, wherein information is acquired. A falsification detection method for determining whether or not a print content of a printed matter has been falsified:
A modified image creation step of creating a plurality of modified images in which different modifications are made to the scanned image by modifying the scanned image created by reading a printed material on which the original image is printed;
A feature quantity measuring step of measuring each feature quantity of the plurality of changed images;
A feature amount difference calculating step for obtaining a difference value between the feature amount of the original image and the feature amount of each changed image;
A determination step of determining whether or not the read image has been falsified based on a value obtained by the feature amount difference calculation unit;
A tamper detection method characterized by including:

Images