JP2010074242A - Image generating apparatus, image generating method, computer executable program, and computer readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image generating apparatus capable of guaranteeing the genuineness of an image comprising a plurality of layers highly accurately, and identifying a tampered part if the image is tampered. <P>SOLUTION: As for an MRC (Mixed Raster Contents) system image comprising a plurality of layers containing a mask layer, layers other than the mask layer are selected out of the plurality of layers as embedded layers. One or a plurality of layers other than the embedded layers are selected as hash value extraction layers. A hash value is computed for every block of the selected hash value extraction layers. The computed hash value is embedded at a corresponding block in the embedded layer, as a digital watermark. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image generation apparatus, an image generation method, a computer-executable program, and a computer-readable recording medium, and more particularly, embeds a digital watermark in an image composed of a plurality of layers including a mask layer. The present invention relates to an image generation apparatus, an image generation method, a computer-executable program, and a computer-readable recording medium.

  Generally, a document is composed of a mixture of characters / line drawings and images. In order to efficiently encode such mixed documents, the characters / line drawings and image portions are separated into different encodings. Is effective. FIG. 17 is a diagram for explaining an MRC (Mixed Raster Contents) model. For example, as shown in FIG. 17, a method called an MRC (Mixed Raster Contents) model is used to convert a document of one page into, for example, character color information (foreground), character area information (mask), and image information (background). This is a method of dividing into layers and performing encoding for each layer.

  The reason for adopting the MRC model is to prevent deterioration of characters and line drawings at a high compression rate. This is because the JPEG algorithm, which is one of the typical compression methods for still images, performs quantization, which is an irreversible process, so that the decompressed image becomes an image with a deteriorated image quality compared to the image data before compression. Usually, a quantization table is used in which the quantization step becomes coarser as the high frequency component increases in order to increase the compression ratio. For this reason, particularly in a portion where there is an edge containing a lot of high-frequency components such as characters and line drawings, the image quality is significantly deteriorated and the compression rate is also reduced. Here, by independently compressing the character and line drawing parts by a reversible compression method or the like, it is possible to increase the compression rate and suppress deterioration in image quality.

  There are various configurations of such MRC. For example, the MRC configuration example 1 in FIG. 18 is configured by superimposing a plurality of character images on one background. This character image consists of a combination of character shape and character color information, and the color in the character shape is a single color.

  In the MRC configuration example 2 in FIG. 19, a mask image showing a plurality of character shapes and a foreground layer that gives color information in the character shapes are superimposed on one background. In the superimposed image, the pixel value of the foreground layer is used for the pixel whose mask image representing the character shape is 1 and the pixel value of the background layer is used for the pixel whose mask image is 0. In FIG. 3, black characters and color characters are separated and configured with a total of four layers, but the basic concept is the same with a three-layer configuration that does not separate black characters.

  By the way, in recent years, with increasing security needs, a technology for preventing falsification detection and information leakage has been demanded. In order to enable tampering detection, data that is generally created from image data by hashing or the like may be embedded in the low-impact region of the image. As a general method, there are a method of embedding the hash value of the upper bits of the pixel value in the lower bits as a watermark, and a method of embedding the hash value of the low frequency component after frequency conversion in the high frequency component of the conversion coefficient. In consideration of authenticity assurance when an image is divided into a plurality of layers of the MRC method, if the digital watermark of the MRC method is embedded before the division into a plurality of layers, the watermark is generated by a quantization process existing in the division process. Will break, so it is necessary to embed a watermark in the image of the divided layers.

  For example, in Patent Document 1, a digital content is divided into a low human sensitivity (low influence area) and a high part (high influence area), the low influence area is encrypted with a public key, and the hash of the copyright information and the high influence area is obtained. A technique is disclosed in which a value is embedded as a digital watermark, decrypted with a secret key, and then a high-impact region and a low-impact region in which the watermark is embedded are combined to generate a watermarked image.

  Since Patent Document 1 is intended for a single image, there is no problem when an electronic watermark is embedded in an original image. However, in the case of an MRC image, the image is divided into a character area, a background area, and a foreground area. When different encoding is performed, for example, an image that is meaningful to human beings is formed in both the character layer and the background layer. Therefore, in the case of an MRC image, each layer is an independent image. Therefore, in the method of Patent Document 1, a digital watermark is embedded in each layer image independently. In this case, for example, there is a problem that the image data of the character mask layer is exposed to an attack that completely replaces the image data with the digital watermark embedded. When subjected to such an attack, if the character mask alone is viewed, the correct digital watermark is embedded, so that it is not detected that the character has been tampered with. However, if the character mask is replaced, serious information tampering occurs. Further, in Patent Document 1, there is a problem that a place where tampering has been performed cannot be specified in order to embed an electronic watermark in data obtained by concatenating bit strings of bit layers.

  In Patent Document 2, a predetermined multi-layer method is performed on an input image to generate a plurality of partial image (layer image) data, which are encoded to generate a partial image (layer image) code. These are put together to generate an output code. Furthermore, a technique for performing an electronic signature with a predetermined granularity on a partial image (layer image) code included in an output code is disclosed.

  However, in Patent Document 2, it is possible to detect the information of the copyright holder and the presence or absence of tampering, but there is a problem that the location of tampering cannot be specified.

JP-A-2005-39686 JP 2007-81596 A

  An object of the present invention is to provide an image generation apparatus and an image generation method capable of guaranteeing the authenticity of an image composed of a plurality of layers with high accuracy and specifying the location of the tampering when tampered. Another object is to provide a computer-executable program and a computer-readable recording medium.

  In order to solve the above-described problems and achieve the object, the present invention selects a layer other than the mask layer as an embedding layer in which a digital watermark is embedded, among images composed of a plurality of layers including a mask layer. An embedding layer selection unit; a generation unit that generates a digital watermark based on a pixel value of an arbitrary extraction region of one or more layers other than the embedding layer; an embedding layer selected by the embedding layer selection unit; And embedding means for embedding the digital watermark generated by the digital watermark generation means in an embedding area corresponding to the extraction area.

  In order to solve the above-described problems and achieve the object, the present invention provides a layer other than the mask layer as an embedded layer in which a digital watermark is embedded, among images composed of a plurality of layers including a mask layer. An embedding layer selection step to be selected, a generation step for generating a digital watermark based on a pixel value of an arbitrary extraction region of one or more layers other than the embedding layer, and an embedding selected in the embedding layer selection step And an embedding step of embedding the digital watermark generated in the generating step in an embedding region corresponding to the extraction region of the layer.

  According to the present invention, an embedding layer selection unit that selects a layer other than the mask layer as an embedding layer in which a digital watermark is embedded, among images composed of a plurality of layers including a mask layer, and other than the embedding layer. Generating means for generating a digital watermark based on a pixel value of an arbitrary extraction area of one or a plurality of layers, and in the embedding area corresponding to the extraction area of the embedding layer selected by the embedding layer selection means, And an embedding unit that embeds the digital watermark generated by the watermark generating unit, so that the authenticity of the image composed of a plurality of layers is ensured with high accuracy, and if the image is tampered, the tampering is performed. There is an effect that it is possible to provide an image generation apparatus capable of specifying the location.

Hereinafter, a preferred embodiment when an image generation apparatus, an image generation method, a computer-executable program, and a computer-readable recording medium according to the present invention are applied to an image processing apparatus will be described in detail. Note that the present invention is not limited to the embodiments. In addition, constituent elements in the following embodiments include those that can be easily assumed by those skilled in the art or that are substantially the same.
(Embodiment 1)

  FIG. 1 is a diagram illustrating a functional configuration example of the image processing apparatus according to the first embodiment. FIG. 2 is a diagram illustrating an example of dividing an input image into a plurality of layers. FIG. 3 is a flowchart for explaining the outline of the overall processing of the image processing apparatus according to the first embodiment. FIG. 4 is a diagram illustrating an example of the encoding process. In the first embodiment, a case will be described in which a hash value is calculated from pixel values of a mask layer subjected to lossless conversion and a watermark is embedded in a foreground or background that is JPEG compressed.

  As shown in FIG. 2, the present invention relates to a technique for embedding a digital watermark in an image composed of a plurality of layers, called an MRC configuration. An image already composed of a plurality of layers may be targeted, and a process for dividing an image into a plurality of layers may be performed in the same apparatus. For example, a paper document is scanned to create an electronic document having a multi-layer structure.

  As shown in FIG. 1, the image processing apparatus 100 according to the first embodiment includes an embedding layer selection unit 100, a hash value calculation unit 102, a frequency conversion unit 103, a digital watermark embedding unit 104, and an encoding unit 105. And a multi-layer documenting unit 106.

  The overall operation of the image processing apparatus 100 of FIG. 1 will be described with reference to FIG. In FIG. 3, the embedding layer selection unit 100 selects a layer other than the mask layer from among the MRC image layers as an embedding layer for embedding a digital watermark, and also selects one or a plurality of hash value extraction layers. (Step S1). The hash value calculation unit 102 calculates a hash value for each block of the selected hash value extraction layer (step S2). The frequency conversion unit 103 performs frequency conversion on the embedding layer, and the digital watermark embedding unit 104 embeds a hash value as a digital watermark in the conversion coefficient of the embedding layer subjected to frequency conversion (step S3). After encoding is performed by the encoding unit 105, the multi-layer documenting unit 106 documents the encoded multi-layer (step S4).

  The operation of each unit of the image processing apparatus 100 in FIG. 1 will be described. The embedding layer selection unit 100 selects an embedding layer for embedding a digital watermark from the layers constituting the MRC for the input image in the MRC format, and outputs the selected embedding layer to the frequency conversion unit 103. The embedding layer selection unit 101 selects a hash value extraction layer for calculating a hash value from the layers constituting the MRC, and outputs the hash value extraction layer to the hash value calculation unit 102 and the multi-layer documenting unit 106. The layer selection unit 101 outputs the other layers (layers other than the embedding layer and the hash value extraction layer) to the multi-layer documenting unit 106. When the embedding layer and the hash value extraction layer are determined, the embedding layer data is directly output to the frequency conversion unit 103, and the hash value extraction layer data is directly output to the hash value calculation unit 102. With this configuration, the embedded layer selection unit 100 is not necessary.

  The hash value calculation unit 102 calculates a hash value to be embedded as a digital watermark from the data of the hash value extraction layer, and outputs the calculated hash value to the digital watermark embedding unit 104. Here, the hash value can be calculated from a single layer data or a plurality of layer data.

  The frequency conversion unit 103 performs an encoding process on the embedding layer, and outputs the converted transform coefficient to the digital watermark embedding unit 104. The encoding process performed here also includes a quantization process depending on circumstances. The type of digital watermark is not a digital watermark embedded in a sample value, but a digital watermark embedded in a frequency component. When the embedded layer is multi-valued, it is often encoded by an encoding method such as JPEG or JPEG2000. In a general encoding process, as shown in FIG. 4, color conversion (S11), frequency conversion (S12), quantization (S13), coefficient shaping (S14), entropy encoding (S15), code stream ( S16) is performed. When irreversible compression is used, information is discarded during the color conversion and quantization process, so the frequency conversion coefficient is embedded in the quantized coefficient rather than embedded in the sample value before encoding. Is good.

  The digital watermark embedding unit 104 processes the conversion coefficient input from the frequency conversion unit 103 by inserting the hash value calculated by the hash value calculation unit 102 and outputs the digital watermarked coefficient to the encoding unit 105. . The encoding unit 105 encodes the digital watermarked coefficient input from the digital watermark embedding unit 104 up to the final code stream, and outputs the digital watermarked layer code to the multi-layer documenting unit 106.

  The multi-layer documenting unit 106 generates and outputs a multi-layer document using all layers of the layer code with digital watermark, the hash value extraction layer, and the non-embedded non-extraction layer. For example, a PDF file can be generated by adding information on the compression method, superimposition information, and page information to the code stream of each layer.

  The hash value calculation processing in the hash value calculation unit 102 will be described in detail with reference to FIGS. FIG. 5 is a flowchart for explaining a hash value calculation process in the hash value calculation unit 102, and FIG. 6 is a diagram for explaining an example of a block corresponding to each layer. FIG. 7 is a diagram for explaining block division of the character mask layer, and FIG. 8 is a diagram for explaining generation of input data from the divided blocks.

  A hash value calculation process performed by the hash value calculation unit 102 will be described with reference to FIGS. As described above, the layer used for calculating the hash value may be single or plural. For example, when a hash value is calculated from two layers of a black character mask image and a color character mask image, it is desirable to calculate the hash value after combining the pixel values of the respective layers. The hash values may be calculated for each layer and then combined, but in this case, the hash value bits increase.

  In FIG. 5, first, the layer is divided into blocks of the same size (step S21). Here, the same size block means that the same size is obtained when the layers are combined. In the MRC configuration, the resolution of each layer may be different. As shown in FIG. 6, for example, the background layer may have a resolution of 100 dpi, the character mask has a resolution of 200 dpi, and the foreground layer has a resolution of 50 dpi. It is. In this case, if the character mask layer is divided into blocks of size 64 * 64, the background layer has a block size of 32 * 32 and the foreground layer has a block size of 16 * 16. In this way, when a block is divided and each layer is combined, the hash value of a layer is embedded in another layer by making it correspond to the same position so that it overlaps the same size. Can be detected in units of blocks.

  Next, a pixel value is extracted from each block (step S22). In FIG. 7, for example, when calculating a hash value from a block divided into L * L size in a monochrome binary character mask layer, L * L bit data is obtained by concatenating the pixel values in the block. Therefore, a hash value is calculated from this L * L bit data. Similarly, the tampering detection requires processing for calculating a hash value, and therefore, when using a pixel value, it is desirable to calculate from layer data using lossless compression.

  Subsequently, the bits used for calculating the hash value are extracted from the pixel value (step S23). For example, SHA-1 (Secure Hash Algorithm) which is one of hash functions generates a 160-bit hash value from data having an arbitrary length less than 64 bits of “2”. In FIG. 8, assuming that the block size is L * L, a bit string of L * L bits can be generated by scanning all the pixel values in the block in a fixed order, which becomes an input of a hash function. Here, when the hash value is calculated from the multi-value data, the number of bits may become too large if the values of the pixels are connected as they are. In such a case, the number of input bits can be reduced by using the upper few bits of the pixel value. For example, if the block size is L * L and the pixel value is 8 bits, a bit string of total L * L * 6 bits can be generated by concatenating the upper 6 bits of each pixel.

  Next, the bits extracted from each layer are synthesized (step S24). Here, when a hash value is calculated using data of a plurality of layers, data obtained by combining the bit strings extracted from each layer is used as an input of the hash function. There are various combining methods. For example, when the number of bits is small, they may be connected as they are, and when the number of bits is large, a new bit string may be generated by thinning out the connected bit strings.

  A hash value is calculated based on the synthesized data (step S25). Specifically, the hash value is calculated using the generated bit string as the input of the hash function.

  The digital watermark embedding process in the digital watermark embedding unit 104 will be described in detail with reference to FIGS. 9 is a flowchart for explaining the digital watermark embedding process of the digital watermark embedding unit 104, FIG. 10 is a diagram showing an example of DCT coefficients after quantization, and FIG. 11 is an example of embedding digital watermarks in a plurality of DCT transform blocks. FIG.

  The digital watermark embedding process of the digital watermark embedding unit 104 will be described with reference to FIGS. 10 and 11 according to the flowchart of FIG. In FIG. 9, first, the layer is divided into blocks of the same size (step S31). Similarly to step S21 in FIG. 5, the block in which the digital watermark is embedded is also divided so that the blocks are aligned at the same position and the same size when the layers are combined. When the coding method of the embedding layer is a coding method using block transformation represented by DCT transformation, the block size of hash value extraction and watermark embedding is determined in accordance with the block size of block transformation. Is desirable. Here, a case where a watermark is embedded in a layer subjected to JPEG compression will be described as an example. In this case, it is desirable to select a size of “8” or “multiple of 8” for the vertical and horizontal directions. This is because the encoding and watermark embedding processes can be performed independently for each block.

  Thereafter, frequency conversion is performed for each block (step S32). Here, frequency conversion is performed for each block used in the encoding method of the embedded layer. For example, when compressing with JPEG, DCT conversion is performed for each 8 * 8 block, and similarly 8 * 8 conversion coefficients are obtained. For details on JPEG encoding, see ISO / IEC IS 10918 | ITU-T Recommendation T. 81, detailed description thereof is omitted.

  Next, the transform coefficient is quantized using a predetermined quantization table (step S33). The extracted hash value is embedded as a digital watermark in the quantized (DCT) transform coefficient (step S34). In the example of the DCT coefficients after quantization shown in FIG. 10, the upper left coefficient is a DC coefficient, and the other AC coefficients are arranged in the lower right side to indicate higher frequency components. Since humans have the property that high frequency components are less perceptible, they are embedded in the lower right frequency component. Further, in order to suppress degradation of image quality as much as possible, a hash value is embedded in the lower bits of the DCT coefficient. Embedding is performed by replacing a predetermined lower bit of the DCT coefficient with an embedded hash value. For example, when 16 * 16 is used as a block size for hash value calculation, as shown in FIG. 11, embedding is performed on four blocks of DCT conversion. It embeds by dividing into high frequency components of AC coefficients of four DCT transform blocks. The wavy line in the figure shows the area where the hash value is embedded. Image quality degradation can be suppressed by embedding each of the four blocks in four coefficients rather than embedding in 16 coefficients in one block. As described above, the watermark embedding method is determined in consideration of the calculated number of hash bits and the granularity of an image to be detected for alteration.

  The DCT coefficient is scanned (step S35). When the DCT coefficient scan order is JPEG, a zigzag scan from the upper left to the lower right is performed. Next, the coefficients arranged in the scan order are entropy encoded (step S36). In the case of JPEG, encoding is performed by either Huffman encoding or arithmetic encoding. Since information is not lost during the entropy encoding process, the coefficient obtained immediately after entropy decoding is a coefficient after embedding a digital watermark. Therefore, the hash value can be extracted from the low-order bits of the embedded high-frequency component and verified.

  Information such as the image size, the number of pixel bits, and the quantization table is added as a header to generate a code stream (step S37).

  The falsification of an image embedded with a digital watermark by the above method can be detected by the following procedure. FIG. 12 is a flowchart for explaining a tampering detection procedure. The processing of the following flowchart can be implemented in an image processing apparatus by software or hardware.

  First, the code stream of each layer of the image is extracted (step S41). The embedded layer is entropy decoded (step S42). Each block of the hash extraction layer is decrypted to calculate a hash H1 (step S43). A hash H2 is extracted from the coefficient of each block obtained by entropy decoding with respect to the embedded layer (step S44). Further, the hash H1 and the hash H2 are compared for each block, and if they are different, it is determined that there is tampering (step S45). In this way, different blocks of the hash H1 and the hash H2 can be specified as the falsification position.

  As described above, according to the first embodiment, a layer other than the mask layer is selected as an embedding layer among the plurality of layers for the MRC scheme image including a plurality of layers including the mask layer, and A hash value extraction layer is selected as a layer other than the embedding layer, a hash value is calculated for each block of the selected hash value extraction layer, and the calculated hash value is embedded as a digital watermark in a corresponding block of the embedding layer Therefore, the falsification detection function is not independent for each layer image, and is robust against an attack in which an arbitrary layer image is replaced with another data among the divided layers and combined. Further, by embedding a digital watermark in an embedding area corresponding to an area for which a hash value has been calculated, it becomes possible to identify a tampered location.

(Embodiment 2)
An image processing apparatus according to Embodiment 2 will be described. The image processing apparatus according to the second embodiment embeds a hash value calculated from the mask layer in the image processing apparatus according to the first embodiment, from among the foreground and background layers, a layer having a small number of pixels referred to at the time of synthesis. It is the structure selected as a layer.

  By the way, in many cases, the number of pixels in which the foreground layer representing characters in the mask layer is used is significantly smaller than the number of pixels in which the background layer representing the pattern is used. In such a situation, since many pixel values of the foreground layer are not used, the image quality is hardly affected by the watermark embedding. Therefore, if a layer with few pixels that are referred to when combining layers is selected as an embedding destination layer, a higher quality MRC image can be generated without changing the function of tampering detection.

  When the image is composed of three layers of a mask layer, a foreground layer, and a background layer, the foreground layer is referred to in the binary image of the mask layer in the process of selecting the embedded layer in step S1 of FIG. Is compared with the number of pixels indicating that the background layer is referred to, and a layer having a small number is selected as an embedded layer. For example, the MRC configuration is foreground layer + mask layer + background layer, and layer synthesis is performed by overlaying the foreground layer image from the top only on the portion of the background layer image where the pixel value of the mask layer is “1”. In this case, when the total number of pixels having the pixel value “0” and the pixel having the pixel value “1” is smaller than the total number of pixels having the pixel value “1”, the foreground layer image Choose to embed hash values in the data. On the other hand, when the number of pixels having a pixel value “0” is small, it is selected to embed a hash value in the image data of the background layer.

  As described above, according to the second embodiment, since the embedding layer is the layer representing the least number of pixels that are referred to when combining the layers among the layers representing the non-mask image, tamper detection is performed. It is possible to generate an MRC image with higher image quality without changing the function.

(Embodiment 3)
An image processing apparatus according to Embodiment 3 will be described. FIG. 13 is a diagram for explaining the image processing apparatus according to the third embodiment, and is a diagram for explaining a case where an embedding layer is selected based on mask layer data for each region (block). . The image processing apparatus according to the third embodiment has a configuration in which, in the image processing apparatus according to the first embodiment, for each hash extraction region, a layer having a small pixel value referred to during layer composition is selected as an embedded layer.

  In FIG. 13, as in the first embodiment, when all layers are combined, they are divided into a plurality of blocks so that they have the same size and the same position. For every block, the hash value calculation and the embedding layer determination are repeated for each block. For each block, the total number of binary values (0, 1) included in the corresponding block of the mask layer image representing the character shape is counted, and the total number of values referring to the pixel values of the background layer at the time of synthesis is If there are many, the hash value calculated from the image of the mask layer is embedded as a digital watermark in the foreground layer.

  According to the third embodiment, among the non-mask layers, the layer with the smallest number of pixels referred to during layer synthesis is selected as the embedded layer for each block of the mask layer, so the alteration detection function is changed. Therefore, it is possible to generate a higher quality MRC image.

(Embodiment 4)
An image processing apparatus according to Embodiment 4 will be described. The image processing apparatus according to the fourth embodiment has a configuration in which the hash value is calculated from the background layer and embedded in the foreground layer in the image processing apparatus according to the first embodiment.

  In an MRC-type image composed of a mask layer, a foreground layer, and a background layer, a case is considered in which a binary image is reversibly compressed in the mask layer, and a multi-valued image is irreversibly compressed in the foreground layer and the background layer. Since the background layer performs irreversible compression, if the hash value is calculated from the pixel value, the pixel value has changed once compressed and decoded, and the hash value calculated in the tamper detection phase even if it has not been altered Will change from the embedding phase.

  Therefore, in the fourth embodiment, when a hash value is calculated from the background layer, the pixel value is not used as it is, but the quantized DCT coefficient is used. Since the DCT coefficient after quantization has significant data in the low frequency component, the low frequency component data can be used as the input of the hash function. The process of embedding the calculated hash value in the embedding layer is the same as in the first embodiment.

  According to the fourth embodiment, since the hash value used for the digital watermark is calculated from the value obtained by frequency-converting the pixels included in an arbitrary extraction area of the hash value extraction layer, the background layer is irreversible. When compression is performed and the background layer is used as a hash value extraction layer, it is possible to prevent the hash value after decoding from changing from that in the embedding phase.

(Embodiment 5)
An image processing apparatus according to Embodiment 5 will be described. FIG. 14 is a diagram for explaining wavelet transform in JPEG2000. FIG. 15 is a diagram illustrating the positions of wavelet coefficients corresponding to mask-off. FIG. 16 shows a bit plane structure.

  The image processing apparatus according to the fifth embodiment is configured to embed a digital watermark at a location corresponding to a mask-off pixel in a layer compressed by an encoding method using wavelet transform in the image processing apparatus of the first embodiment It is.

  JPEG2000 is an encoding method using discrete wavelet transform, and the flow of the encoding process is generally the same as the flow of the general encoding process described above. The difference between the DCT transform and the wavelet transform is that the transform coefficient obtained by the DCT transform is a component centered only on the frequency in the 8 * 8 block, whereas the wavelet transform has a spatial position in addition to the frequency. It is a point that is a component with an axis. FIG. 14 shows the structure of transform coefficients obtained when wavelet transform is performed, and the upper left part divided by rectangles at each resolution level represents a low frequency component, and the lower right part represents a high frequency component. Further, the positions of the transform coefficients in the rectangle are arranged corresponding to the positions of the original image.

  Therefore, the mask is turned off, and in the watermark embedding layer, the wavelet transform coefficient corresponding to the pixel position that is not referred to at the time of layer synthesis can be specified. That is, it is possible to identify a transform coefficient that is not used for layer synthesis from the mask layer value, and directly embed a watermark only in the transform coefficient. Thereby, it is possible to suppress degradation of image quality due to digital watermark embedding. In the example shown in FIG. 15, the halftone dot portion indicates the position of the wavelet coefficient corresponding to mask-off.

  In addition, when digital watermark embedding is performed, since the image quality degradation is smaller when the digital watermark is embedded in the high frequency component, for example, it may be embedded in the HH component which is the lower right rectangular portion. Furthermore, since the image quality degradation can be reduced as the LSB in the bit plane structure shown in FIG. 16 is closer, it is desirable to embed the hash value bits from the LSB.

  According to the fifth embodiment, since the digital watermark is embedded in the conversion coefficient of the embedding layer corresponding to the pixel indicating the mask-off of the mask layer, it is possible to suppress deterioration in image quality due to the embedding of the digital watermark.

(program)
Note that the image processing apparatus according to the present invention can be applied to a system including a plurality of devices (for example, a host computer, an interface device, a scanner, a printer, and the like). ) May be applied.

  Another object of the present invention is to supply a recording medium recording a program code of software for realizing the functions of the above-described image processing apparatus to the system or apparatus, and the computer of the system or apparatus (or CPU, MPU, It can also be achieved by the DSP) executing the program code stored in the recording medium. In this case, the program code read from the recording medium itself realizes the functions of the image processing apparatus described above, and the program code or the recording medium storing the program constitutes the present invention. Recording media for supplying the program code include FD, hard disk, optical disk, magneto-optical disk, CD-ROM, CD-R, magnetic tape, non-volatile memory, optical recording medium such as ROM, magnetic recording medium, optical Magnetic recording media and semiconductor recording media can be used.

  Further, by executing the program code read by the computer, not only the functions of the image processing apparatus described above are realized, but also an OS (operating system) operating on the computer based on the instruction of the program code. However, it goes without saying that a case where the function of the image processing apparatus described above is realized by performing part or all of the actual processing.

  In addition, after the program code read from the recording medium is written in a memory provided in a function expansion board inserted in the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing, and the above-described functions of the image processing apparatus are realized by the processing.

2 is a diagram illustrating a functional configuration example of the image processing apparatus according to Embodiment 1. FIG. It is a figure which shows the example which divides | segments an input image into several layers. 3 is a flowchart for explaining an outline of overall processing of the image processing apparatus according to the first embodiment; It is a figure which shows an example of an encoding process. It is a flowchart for demonstrating the calculation process of the hash value in a hash value calculation part. It is a figure for demonstrating an example of the block to which each layer respond | corresponds. It is a figure for demonstrating the block division of a character mask layer. It is a figure for demonstrating the production | generation of input data from the divided | segmented block. It is a flowchart for demonstrating the embedding process of the digital watermark of a digital watermark embedding part. It is a figure which shows the example of DCT coefficient after quantization. It is a figure which shows the example which embeds a digital watermark in several DCT conversion blocks. It is a flowchart for demonstrating a tampering detection process. It is a figure for demonstrating the case where an embedding layer is selected based on the data of a mask layer for every area | region (block). It is a figure for demonstrating the wavelet transformation in JPEG2000. It is a figure which shows the position of the wavelet coefficient corresponding to mask-off. It is a figure which shows a bit plane structure. It is a figure for demonstrating a MRC (MixedRasterContents) model. It is a figure for demonstrating the structural example 1 of MRC. It is a figure for demonstrating the structural example 2 of MRC.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Image processing apparatus 101 Embedding layer selection part 102 Frequency conversion part 103 Digital watermark embedding part 104 Encoding part 105 Multilayer document part

Claims (14)

An embedding layer selection means for selecting a layer other than the mask layer as an embedding layer for embedding a digital watermark among images composed of a plurality of layers including a mask layer;
Generating means for generating a digital watermark based on a pixel value of an arbitrary extraction region of one or more layers other than the embedded layer;
Embedding means for embedding the digital watermark generated by the generating means in an embedded area corresponding to the extraction area of the embedded layer selected by the embedded layer selecting means;
An image generation apparatus comprising:   The image generation apparatus according to claim 1, wherein the embedding layer selection unit selects the embedding layer based on data of the mask layer.   The image generation apparatus according to claim 2, wherein the embedded layer selecting unit selects, as the embedded layer, a layer having the smallest number of pixels referred to when combining the layers.   The image generation apparatus according to claim 1, wherein the embedding layer selection unit selects the embedding layer for each extraction region.   The image generation apparatus according to claim 4, wherein the embedded layer selection unit selects the embedded layer based on the data of the mask layer for each of the extraction regions.   The image generation apparatus according to claim 5, wherein the embedded layer selection unit selects, as the embedded layer, a layer having the smallest number of pixels referred to when combining layers for each extraction region. The image generation apparatus according to claim 1, wherein the embedding unit performs digital watermark embedding on a transform coefficient of the embedding layer corresponding to a pixel indicating mask-off.   The image generation apparatus according to claim 1, wherein the embedded layer is encoded by an encoding method using DCT transform.   The image generation apparatus according to claim 1, wherein the embedded layer is encoded by an encoding method using wavelet transform.   The hash value used for the digital watermark is calculated from a value obtained by frequency-converting pixels included in an arbitrary extraction region of the extraction source layer. The image generating apparatus described.   10. The hash value used for the digital watermark is a hash value calculated from data obtained from layer data and data obtained by combining a time stamp. The image generating apparatus described in 1. An embedding layer selection step of selecting a layer other than the mask layer as an embedding layer for embedding a digital watermark among images composed of a plurality of layers including a mask layer;
A generation step of generating a digital watermark based on a pixel value of an arbitrary extraction region of one or more layers other than the embedded layer;
An embedding step of embedding the digital watermark generated in the generation step in an embedding region corresponding to the extraction region of the embedding layer selected in the embedding layer selection step;
An image generation method comprising:   A computer-executable program for causing a computer to function as each means of the invention according to any one of claims 1 to 11.   A computer-readable recording medium storing the computer-executable program according to claim 13.

Images