DCT transform digital watermark embedding method based on human visual masking

Technical Field

The content of the invention in human visual masking characteristic on the basis of DCT transformation to, which belongs to the technical field of Image watermark embedding, in particular to the protection of the copyright of the original picture, also belongs to the technical field of information hiding.

Background Art

Today's Internet communication has become a living in a part of the can be erased, multimedia as an important medium for the exchange of information is playing an increasingly important role, various forms of digital products, in an endless stream. Then more and more people select network express their outstanding digital works or carry on , but all have the opposite side, a large number of piracy works is also increased, for example do not have any copyright of the Image, the television and the like taken by people using various copies, copyright is not authorized works will be perpetrators dissemination or proliferation of prohibited sales, these are serious harm to the copyright owner of Intellectual property rights, secondly, due to the characteristic of the digital works rich, one of them is the digital product copy does not affect the quality of the work itself, and the fabrication cost is very low, so vulnerable to renovate and re-use. We need to protect these works the legal rights and interests of the owner, the industry can more high-speed development. A solution to this problem is the development of a center of gravity of the information hiding technique, the digital watermark is a core technology in the technical field.

Application number 201210165687.5 name is "based on the visual attention mechanism of digital Image watermarking method" referred to in the Patent of digital watermark system of two major characteristics: robustness and visibility, wherein the invisibility depends on the human visual characteristics, in this Patent through the visual attention to the characteristics of the target watermark embedded from the outset of the prominent area of the picture, thus the extracted watermark has better robustness, but visual attention can be composed of two kinds of patterns of: one is the objective content-driven bottom-up attention model; the other type is a subjective command guidance attention model, any kind of the are too dependent on the psychological activities, will inevitably to judge the impact of the topographic.

Application number 200910021564.2 name is "based on the DCT algorithm digital Image watermark embedding method" Patent algorithm in black-and-white Image is embedded into the texture area of the original Image, however did not mention why to be embedded to the original picture of the texture area, and in the embedding process rather than by directly utilizing the water in the black-and-white Image printing element some non-0 that 1 the characteristics of the embedding.

Application number 201410479555.9 name is "a based on the Image content of the digital watermarking algorithm" Patent realizes a high capacity of the embedding algorithm, but only the original picture for the protection of the copyright on as much as possible and it is not necessary to do so embedded, the need to ensure that the pit offensive high, the Patent to intermediate frequency coefficient in the DCT is the extraction of the 22 a multi-, also to take into consideration the Image texture, edge and a sliding three characteristics, in practice somewhat cumbersome.

As shown in the foregoing, based on the human visual masking characteristic picture in the texture area has greater brightness and direction change, the area information of the human eye to resolution deterioration, namely texture masking, we to the host Image texture region can be successfully embed the watermark Image to the human eye can not recognize the, also in left in the original Image of the protection of copyright sign, at the same time we select the position of the watermark in the Image on the binary, this let US the algorithm rules more simply in the completion of the embedding of the watermark with the original Image of copyright protection.

Content of the invention

The content of the invention is for the protection of the copyright of the original picture to provide a based on human visual conceal the DCT transform of the digital Image watermark embedding method, as far as possible to the same time of the algorithm is simple and effective, and the embedding of the watermark robustness of the picture to a certain extent.

To achieve the above object, the technical scheme comprises the following steps:

1) To be embedded on the conversion of the binary one-dimensional sequence, according to the binary images of the m * n from left to right, from top to bottom scanning expressed as 0, 1 sequence;

2) The original Image by the 8 * 8 block, the adoption of the following formula to calculate the variance of the variance, and the variance from these according to the sequential arrangement of the large to small, find the variance of the largest m * n Image block;

Wherein m each block like point in the average gradation, f (i, j) corresponding to each gray level is.

3) These m * n Image block of DCT transform is carried out, the following formula;

In the above formula: when the u=0 or v=0 when, When the u, v=1, 2, ... N - 1 when, .

4) The fact that the DCT block of a m * n in each block, the two position of the band in the comparison of the magnitude relation of DCT coefficient, at the same time as the difference processing, the original difference larger coefficient of that half of the reduce the processing, the original difference smaller coefficient of half of the increasing treatment, in order to confirm the block embedded in a water printing point "0" or a "1";

5) From the watermark Image converted one-dimensional 0, 1 sequence embedded in the original Image, finally through the inverse transform to the DCT, forming the watermark Image, DCT inverse transform formula is as follows.

With the other watermark embedding method is compared, there are also certain anti-offensive, can resist a white Gaussian noise, shearing, JPEG compression and so on different extent different types of attacks, but the embedding process more simple, for the protection of the original picture copyright have the advantages of high efficiency and fast.

Description of drawings

Figure 1 is the flow diagram of the content of the invention;

Figure 2 is the original plan of the content of the invention in an embodiment of sheet;

Figure 3 is the of the content of the invention in one embodiment of embedded to sheet;

Figure 4 is the diagram of the content of the invention in an embodiment of the watermark has been embedded into the sheet.

Specific embodiment:

The following of the Figure embodiment of the invention the content of the further detailed description.

The implementation of the example one

See Figure 1 of the process, will be a 32 * 32 of the Chinese character "dawn" as the binary watermark embedded in the Image to the 512 * 512 of the original Image in the classical lina

First of all, to be embedded into the Chinese character "dawn" binary to one-dimensional sequence of the conversion, the 32 * 32 according to the binary from left to right, from top to bottom scanning expressed as 0, 1 sequence; then the original Image on the lina 8 * 8 block, by calculating the variance, and the variance from these according to the sequential arrangement of the large to small, to find the maximum variance of the 32 * 32 Image block; these 32 * 32 Image block is DCT transformation, wherein the selected in these Image block (4, 3) and (5, 2) two position of the dct coefficient, the coefficient of the two difference processing, the original difference larger coefficient of that half of the reduce the processing, the original difference smaller coefficient of half of the increasing treatment, and the like to the point 1 of the ensure that the former coefficient is smaller than the latter, as is the point 0 of the ensure that the former coefficient is larger than the latter; after the watermark Image converted from the one-dimensional 0, 1 sequence of the original Image in the lina, finally through the inverse transform to the DCT, forming the lina of the watermark Image.