JP3754845B2 - Image processing apparatus and method, and storage medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing apparatus, method, and storage medium, and more particularly, to an image processing apparatus, method, and storage medium for embedding information in digital image data so as to be invisible or extracting the embedded information.
[0002]
[Prior art]
Conventionally, various methods have been developed for digital watermark technology as a method for protecting the copyright of digital contents. This method embeds digital content handling information such as the name of the copyright holder and the purchaser's ID in the digital information of the image so that it cannot be seen by the eyes, and can track unauthorized use by illegal copying As a security and copyright protection technology in electronic distribution, it has recently attracted attention. Furthermore, a digital watermark technique has been developed as a means for preventing digital content from being falsified. In this digital watermark technology, various methods have been proposed as data embedding methods. One method is to embed information using a mask pattern. This method repeatedly embeds digitized image data in accordance with a mask pattern. For example, at the positions a, b, c, and d shown in the mask patterns of FIGS. In accordance with the mask pattern arrangement as shown in FIGS. 2 to 4, information is embedded using a quantization error or the like to obtain a composite image.
[0003]
However, in order to improve the accuracy of specifying the tampered part in the composite image obtained in this way, it is necessary to arrange the mask pattern without gaps in the image data as shown in FIGS. Furthermore, in order to improve the detection accuracy of watermark information from an image partially cut out from the synthesized image, a mask pattern arrangement as shown in FIG. 4 is generally desirable. Therefore, in order to satisfy both the accuracy of specifying the falsification location and the improvement of the watermark resistance against clipping, embedding is performed with the mask pattern arrangement as shown in FIG.
[0004]
[Problems to be solved by the invention]
By embedding by such a method, it is possible to improve the accuracy of specifying the tampering location and the resistance to cutting, but there are the following problems.
1) In order to improve tolerance, the smaller the mask pattern size, the better. However, low-frequency noise and block noise are visually noticeable and cause image quality degradation. The amount of information that can be embedded is limited by the number of mask pattern data.
[0005]
For example, when the mask patterns shown in FIGS. 1A to 1D are used, the mask pattern is limited to 4 bits a, b, c, d, and 16 bits at the maximum.
2) In order to improve the image quality, the larger the mask pattern size, the better, but the resistance becomes weaker.
3) Although it is necessary to increase the embedding position in order to improve the accuracy of specifying the tampered portion, low frequency noise and block noise are visually noticeable and image quality deterioration occurs.
[0006]
Therefore, there is a trade-off relationship between the accuracy of specifying the tampered part, the improvement in cut-out resistance, and the image quality. If one is improved, the other is worsened and both cannot be satisfied.
[0007]
The present invention provides an image processing apparatus and method, and a storage medium that can solve such problems, reduce image quality deterioration, and improve the accuracy of specifying a falsification point.
[0008]
[Means for Solving the Problems]
In order to solve this problem, for example, an image processing apparatus of the present invention has the following configuration. That is,
An image processing apparatus for embedding predetermined information in image data,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristics and a generating means for generating the mask pattern;
And embedding means for embedding the predetermined information by modulating image data corresponding to the embedding target position by applying the mask pattern to a part of the image data.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments according to the present invention will be described in detail with reference to the accompanying drawings.
[0010]
FIG. 14 is a block configuration diagram of the information processing apparatus in the embodiment. In the figure, reference numeral 1 denotes a CPU that controls the entire apparatus, and reference numeral 2 denotes a ROM that stores a BIOS and a boot program. Reference numeral 3 denotes a RAM which is used as a work area of the CPU while the OS and various applications are loaded, and an image memory 3a for storing the read image is secured. Reference numeral 4 denotes an external storage device such as a hard disk, in which an OS and various application programs are stored. Further, image data obtained as a result of processing can be stored as a file. Reference numeral 5 denotes a keyboard (including a pointing device such as a mouse), and reference numeral 6 denotes an image scanner that reads a document image as color image data, and is connected by an interface such as SCSI. A display control unit 7 includes a display memory for storing image data to be displayed. Reference numeral 8 denotes a display device that displays a video signal output from the display control unit 7, and is a CRT, a liquid crystal display, or the like. Reference numeral 9 denotes a color printer, which includes, for example, a printer engine having a recording head for discharging ink droplets by thermal energy only for each recording color component. Reference numeral 10 denotes a network communication unit for performing information communication with a network (which may of course be the Internet).
[0011]
In the above configuration, in the embodiment, the color image data read by the image scanner is stored in the image memory 3a, and separately set information is embedded using an electronic watermark technique.
[0012]
FIG. 5 shows a mask pattern of 32 × 32 pixels having a so-called blue noise characteristic composed of middle and high band components not including a low frequency component. Black dots in the figure indicate information embedding positions. Here, the amount of information (number of bits) to be embedded is set to be equal to or less than the number of black spots.
[0013]
The mask pattern having the blue noise characteristic is usually used for binarization of an image. In the present embodiment, information is embedded by using the pattern having the blue noise characteristic. It is possible to perform efficient electronic watermark processing in consideration of human visual characteristics. The pattern configuration of the blue noise mask is disclosed in, for example, Digital Halftoning, Robert Ulichney (1987, Massachusetts Institute of Technology or Japanese Patent No. 2622429).
[0014]
FIG. 6 shows an arrangement of a mask pattern when information is embedded in the entire image data with the mask pattern of FIG. Hereinafter, an embedding and detection method using the arrangement of the mask pattern of FIG. 5 and the mask pattern of FIG. 6 will be described.
[0015]
In embedding, first, as described above, original image data is read from the image scanner into the image memory 3a. Next, information is embedded in the embedding position of the mask pattern in FIG. 5 starting from the upper left of the mask pattern arrangement in FIG. 6, and when completed, the right mask pattern arrangement is embedded. This process is performed to the lower right of the mask pattern arrangement in FIG.
[0016]
FIG. 7 shows how the position of the mask pattern to be embedded moves in the above-described embedding. Information is embedded at the mask pattern embedding position while moving the mask pattern from the left end of the image in the direction of the arrow.
[0017]
FIG. 8 shows a method of embedding in the image data corresponding to the embedding position of the mask pattern in FIG. X is a pixel value, h is a quantization step width, and n is a natural number.
[0018]
As for the embedding rule, when the embedding information is 0, the quantization is performed to the even-numbered requantized data, and when the embedding information is 1, the quantization is performed to the odd-numbered requantized data.
[0019]
In the example shown in the figure, the target pixel value X is between 2n and 2n + 1. However, if the target pixel value X is between 2n-1 and 2n, when 0 is embedded in the position of the target pixel. Since the pixel value of interest is quantized to 2n which is an even number and 2n-1 which is an odd number in the case of 1, the contradiction does not occur.
[0020]
Briefly, it is assumed that the input pixel is represented by 8 bits for each of RGB and the B component is embedded. In this case, the B component of a certain pixel of the input image can take any value of 0, 1,..., 255. However, if the quantization step width is 8, the information is embedded and after quantization. This data can take values of 0, 8, 16, 24, 32. Here, 0, 16, 32,... Correspond to the above 2n, that is, an even number, and 8, 24,... Correspond to 2n + 1, that is, an odd number.
[0021]
When detecting data embedded in the reverse direction, basically, if the quantization step width is an even multiple (0, 16, 32,...), The data embedded in the data is 0. If it is an odd multiple (8, 24...), It can be detected as 1. However, this is for the pixel determined to be the position to be embedded. In the case of other data (for example, 10), at least information is not embedded in the pixel position.
[0022]
Now, image data converted by the embedding method (FIG. 9A), a part of the image data cut out (FIG. 9B), and image data obtained by falsification of the image data in FIG. 9 (FIG. 9). 10 (a)) and image data (FIG. 10 (b)) obtained by cutting and tampering a part of FIG. 9 (a).
[0023]
In this case, as shown in FIG. 11, the upper left corner of each image is used as a starting point, and it is determined whether or not information is embedded by examining the dots shown in black in the mask pattern shown in FIG. At this time, the coordinates of Point (x, y) and the number of determinations in FIG. 11 are recorded.
[0024]
Whether or not information is embedded in the dot is determined by whether or not the value of each pixel masked by the black dot in the mask pattern in FIG. 5 is an integral multiple of the quantization step width. . Then, the number determined to be an integer multiple (hereinafter referred to as the determination number) and Point (x, y) indicating the position of the mask pattern at that time are associated with each other and stored in an appropriate area in the RAM 3. .
[0025]
Thereafter, this process is repeated while moving rightward by one pixel. After reaching the right end, the process moves to a position shifted by one dot downward from the upper left corner of FIG. 11, and the same process is repeated.
[0026]
Thus, when the determination process for the entire target image is completed, the number of determinations is used as a key to sort in descending order. Then, the coordinates of Point (x, y) and the number of determinations, which are the number of determinations equal to or greater than the threshold, are obtained. Here, the amount of information (number of bits) to be embedded may be used as the threshold value. However, if the author name requires q bits, and the information to be embedded is Q (number of black dots in the mask pattern>Q> q) bits, the remaining bits (Q-q) include parity or other appropriate values. It is sufficient to include such a value.
[0027]
FIG. 12 shows the k sets of information obtained in this way (shows that there are k numbers greater than or equal to the threshold).
[0028]
Then, the embedded information is determined for each set of information for the mask pattern embedding position shown in FIG. At this time, determination information corresponding to the embedded positions of each set is recorded.
[0029]
FIG. 13 shows determination information corresponding to the embedded positions of P1 to Pn in each group. As described above, it is determined that “0” is embedded if the pixel data X is an even multiple of the quantization step width h, and “1” is embedded if the pixel data X is an odd multiple.
[0030]
Then, the majority of “0” and “1” between the same embedding positions is determined, and the embedded information is determined.
[0031]
By the way, the tampering part is specified as follows.
[0032]
X and y of point (x, Y) occupying the majority of the majority are values depending on the size of the mask pattern (32 × 32 in the embodiment). More specifically, when there is no falsification, when the upper left corner position of the input image is x = 0 and Y = 0, x and y of point (x, y) are basically x = 32 × i + c1.
y = 32 × j + c2
Here, i, j = 0, 1, 2,..., And c1 and c2 are constants (of course, depending on the input image). Since information is embedded in the entire image, i and j should be continuous.
[0033]
Therefore, if x and y of point (x, y) are in the above relationship, the region at i = 1, 2, 3, 4, 10, 11 when j = 5 is determined as the region to be embedded. In this case, it may be determined that each area indicated by j = 5 and i = 5-9 is falsified.
[0034]
When the falsified location is determined in this way, a message indicating that the falsification has been made and the falsified location in the input image data are displayed on the display device 8 so as to be distinguished from, for example, a non-falsified location. As an example of the distinction display, it may be performed by surrounding the falsified portion with a frame or displaying it with a different color.
[0035]
In the embodiment, the case where the square mask pattern is used as shown in FIG. 6 has been described as the information to be embedded, but the mask pattern may generally have a size of M × N.
[0036]
In the embodiment, the amount of information to be embedded (the number of bits) is limited to the number indicated by the black dots of the mask pattern. However, when storing information exceeding the number of bits, two mask patterns are stored. What is necessary is just to embed information A and B repeatedly straddling.
[0037]
Further, when embedding information, it is possible to further increase the reliability by including the actually required information (number of bits) + the error correction code bits.
[0038]
The above processing is specifically performed by the CPU 1, and the procedure (program) will be described with reference to FIGS. The program is stored in the external storage device 4, and is loaded onto the RAM 3 and executed.
[0039]
FIG. 15 shows the procedure of the information embedding process using the electronic watermark technique.
[0040]
First, in step S1, an image to be embedded is input from the image scanner 6 and developed in the image memory 3a. In step S2, information to be embedded (for example, a copyright owner name) is input from the keyboard 6 or the like, and a mask pattern is loaded from the external storage device 4 in step S3. As shown in FIG. 5, this mask pattern is a middle-high band, that is, a middle-high frequency component pattern.
[0041]
Next, in step S4, 0 is substituted for each of x and y in order to initialize the position where the mask pattern is applied to the image data developed in the image memory 3a.
[0042]
In step S5, the upper left corner of the mask pattern is set to the x and y positions of the image data, and the predetermined dot group in the black dots in the mask pattern depends on the information (bits) to be embedded. To quantize. This process is performed for the number of bits of information to be embedded.
[0043]
When this processing is completed, the process proceeds to step S6, where it is determined whether or not the right end of the image has been reached from the size of the input image, the size of the mask pattern, and the x and y values at that time. If not, the process of step S7 is performed to shift the mask pattern width to the right, and then the process returns to step S5 to repeat the above process.
[0044]
If it is determined that the right end of the image has been reached, x is initialized to 0, and y is increased by the height of the mask pattern. Then, the processes in steps S5 to S8 are repeated until it is determined in step S9 that the padding for one screen has been completed.
[0045]
Thus, when the embedding process for the input image is completed, the image data stored in the image memory 3a is embedded with information by the electronic watermark technique, and the image data is output. In the case of saving, the output destination is the external storage device 4 or on a network (including the Internet).
[0046]
Next, information extraction and tampering determination processing in which an image with information embedded as described above is embedded will be described with reference to the flowchart of FIG.
[0047]
First, in step S21, an image to be determined is input and developed in the image memory 3a. The input source may be one downloaded from the network, one stored in a floppy or the like, and need not be a specific one. In step S22, a mask pattern (see FIG. 5) is loaded from the external storage device 4.
[0048]
Next, in step S23, 0 is substituted for each of x and y in order to initialize the position to which the mask pattern to be determined is applied to the image data developed in the image memory 3a.
[0049]
In step S24, the upper left corner of the mask pattern is set to the x and y positions of the image data, and all pixel values of the input image corresponding to the black dots in the mask pattern are read and embedded in step S25. The number of pixels that may be present is counted, and the counting result is temporarily stored in a RAM or the like. At this time, the values of x and y fitted with the mask pattern at that time are also stored.
[0050]
In step S26, it is determined whether or not the right end of the image has been reached. If not, x is increased by “1”, that is, the position where the mask pattern is applied is shifted by one pixel, and the process returns to step S24. .
[0051]
If it is determined that the right end of the image has been reached, x is initialized to “0” in step S28, and y is increased by “1”. Then, the processes in steps S24 to S28 are repeated until it is determined in step S29 that the processes in steps S24 and S25 have been completed for the entire image.
[0052]
Thus, when the information collection for one screen is completed, the data stored in step S30 is arranged in the order of counting, and the data of a predetermined number or more is validated. In step S31, the embedded information is extracted.
[0053]
In step S32, based on the collected information and the respective x and y values, it is determined whether or not it has been tampered with, and if it has been tampered with, the position is determined. If it is determined that the information has not been tampered with, processing such as displaying the information embedded in step S33, for example, displaying the name of the copyright holder, is performed.
[0054]
If it is determined that the image has been tampered with, a message indicating that the image has been tampered with is displayed, and an error process specifying the tampered location is performed, and the process ends.
[0055]
As described above, according to the present embodiment, the mask pattern used for embedding information is a mask pattern that does not have a low frequency component, and the mask pattern is black, that is, all the positions where the mask pattern is “1”. Information is not embedded, but quantization processing and information embedding are performed in a limited portion, so that there is little influence on image quality, and good image quality can be maintained. In addition, even if tampering occurs, the position can be specified.
[0056]
Of course, the present invention may be applied to a single device or a combination of a plurality of devices.
[0057]
In addition, the present embodiment requires, for example, means for inputting an image (for image scanner, network connection, or hardware such as a floppy disk), and these are usually used by general-purpose information processing devices (personal computers). The processing may be a general-purpose device that can be provided, or a connectable general-purpose device.
[0058]
Therefore, the present invention supplies a storage medium recording software program codes for realizing the functions of the above-described embodiments to a system or apparatus, and the computer (or CPU or MPU) of the system or apparatus stores the storage medium in the storage medium. Needless to say, this can also be achieved by reading and executing the programmed program code.
[0059]
In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the storage medium storing the program code constitutes the present invention.
[0060]
As a storage medium for supplying the program code, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.
[0061]
Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) operating on the computer based on the instruction of the program code. It goes without saying that a case where the function of the above-described embodiment is realized by performing part or all of the actual processing and the processing is included.
[0062]
Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into 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 functions of the above-described embodiments are realized by the processing.
[0063]
【The invention's effect】
As described above, according to the present invention, in the electronic watermark technology, it is possible to reduce image quality degradation and improve the accuracy of specifying a falsification point.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an example of a region to be embedded by an electronic watermark technique.
FIG. 2 is a diagram showing an example of an array of mask patterns for embedding information.
FIG. 3 is a diagram illustrating an example of an array of mask patterns for embedding information.
FIG. 4 is a diagram showing an example of an array of mask patterns for embedding information.
FIG. 5 is a diagram showing an example of a mask pattern in the embodiment.
FIG. 6 is a diagram illustrating a region where information is embedded by a mask pattern in the embodiment.
FIG. 7 is a diagram showing an embedding sequence in the embodiment.
FIG. 8 is a conceptual diagram of information embedding in the embodiment.
FIG. 9 is a diagram illustrating an example of an embedded image.
FIG. 10 is a diagram illustrating an example of a falsified image.
FIG. 11 is a diagram showing a sequence for applying a mask pattern in embedding information extraction processing;
FIG. 12 is a diagram illustrating a storage state of collected information.
FIG. 13 is a diagram for explaining processing for extracting embedded information;
FIG. 14 is a block configuration diagram of an apparatus according to the embodiment.
FIG. 15 is a flowchart showing an embedding process in the embodiment.
FIG. 16 is a flowchart showing embedding information extraction and falsification determination processing in the embodiment;

Claims (10)

An image processing apparatus for embedding predetermined information in image data,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristics and a generating means for generating the mask pattern;
Embedding means for embedding the predetermined information by applying the mask pattern to a part of the image data and modulating the image data corresponding to the embedding target position;
An image processing apparatus comprising: The image processing apparatus according to claim 1, wherein the mask pattern is expressed by binary information. The image processing apparatus according to claim 1, wherein in the mask pattern, an embedding target position and a non-embedding target position within an M × N size are expressed by binary information. The image processing apparatus according to claim 1, wherein the modulation is quantization of image data corresponding to the position to be embedded. 2. The embedding unit embeds the predetermined pattern by repeatedly applying the mask pattern in addition to a part of the image data and modulating the image data corresponding to the embedding target position. An image processing apparatus according to 1. An image processing method for embedding predetermined information in image data,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristic and a generation step to be generated;
An embedding step of applying the mask pattern to a part of the image data and embedding the predetermined information by modulating image data corresponding to the embedding target position;
An image processing method comprising: An image processing program for embedding predetermined information in image data,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristic and a generation step to be generated;
An embedding step of applying the mask pattern to a part of the image data and embedding the predetermined information by modulating image data corresponding to the embedding target position;
A storage medium that stores an image processing program having a readable state from a computer. An image processing apparatus that extracts information from image data in which predetermined information is embedded,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristics and a generating means for generating the mask pattern;
Extracting means for applying the mask pattern to a part of the image data and extracting the predetermined information by detecting a modulation state of the image data corresponding to the position to be embedded;
An image processing apparatus comprising: An image processing method for extracting information from image data in which predetermined information is embedded,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristic and a generation step to be generated;
An extraction step of applying the mask pattern to a part of the image data and extracting the predetermined information by detecting a modulation state of the image data corresponding to the position to be embedded;
An image processing method comprising: An image processing program for extracting information from image data in which predetermined information is embedded,
A mask pattern specifying an embedding target position in the M × N size according to the blue noise characteristic and a generation step to be generated;
An extraction step of applying the mask pattern to a part of the image data and extracting the predetermined information by detecting a modulation state of the image data corresponding to the position to be embedded;
A storage medium that stores an image processing program having a readable state from a computer.

Images