CN110232651A - A kind of reversible information hidden method based on block prediction and difference histogram shifting - Google Patents

Abstract

The invention discloses a reversible information hiding method based on block prediction and difference histogram translation, which belongs to the field of digital watermarking and information hiding. The method of the invention first divides the image into 4*4 image blocks, and then performs three-neighborhood Blocking, every four pixels as a group, select a pixel and its surrounding three pixels, use the surrounding three pixels to generate a predicted value, use the predicted value to compare with the selected pixel, generate a difference, and the difference is carried down Rounding to an integer, four difference values are generated in one image block, and then a difference value histogram is generated. The invention improves the embedding capacity of the watermark, simplifies the algorithm steps on the premise of maintaining the visual effect of the image, and only needs to be divided into blocks before embedding and extracting the watermark; the method can not only realize the reversible watermark, but also make the watermark invisible Better, the experiment is simple and easy, the PSNR of the image after embedding the watermark is high, the visual effect of the image is good, and it has practical value.

Description

A Reversible Information Hiding Method Based on Block Prediction and Difference Histogram Translation

technical field

The invention belongs to the field of digital watermarking and information hiding, in particular to a reversible information hiding method based on block prediction and difference histogram translation.

Background technique

With the improvement of people's security awareness and intellectual property awareness, people pay more and more attention to the security of their own information, and the use of digital watermarking technology is also increasing. Image digital watermarking technology can be divided into reversible and irreversible watermarking technology according to whether the embedded watermark image can be restored. In order to protect the integrity of images, reversible watermarking is used more and more. Reversible watermarking, also known as lossless watermarking, is a fragile digital watermarking. It can clearly reflect whether the signal has been tampered with, and can restore the original image losslessly after extracting the watermark, which is suitable for the authentication of image integrity. Therefore, reversible image watermarking technology is mainly used in special fields such as medicine, military affairs, and law to preserve and store pictures to ensure that important image information is not lost.

The watermark embedding method of the document "NI Z, SHI Y Q, ANSARI N, et al. Reversible data hiding" (IEEE Transon Circuits and Systems for Video Technology-nology, 2006, 16: (3) 354-362.), the algorithm is easy to implement , the capacity of the embedding also mainly depends on the peak point, but when used in an image with equalized pixel values, the peak point is not prominent, so the effect of this algorithm is not good. The literature "THODI D M, RODRUGUEZ J J. Expansionembedding techniques for reversible watermarking" (IEEE Trans-actions on Image Processing, 2007, 16(3): 721-730) proposes that the predicted value can be generated by calculating adjacent pixels, and the predicted value can be compared with The difference between the original value and the embedding of the watermark depends on a large number of surrounding pixels, accounting for the embedding capacity.

Contents of the invention

Aiming at the above-mentioned technical problems in the prior art, the present invention proposes a reversible information hiding method based on block prediction and difference histogram translation, which has a reasonable design, overcomes the deficiencies of the prior art, and has good effects.

In order to achieve the above object, the present invention adopts the following technical solutions:

A reversible information hiding method based on block prediction and difference histogram translation, comprising the following steps:

Step 1: Set the original image as an M*N grayscale image, and divide the original image into 4*4 image blocks, denoted as B _k , where k=1,2,3...[M/4]* [N/4];

Step 2: Each 4*4 image block B _k is divided into blocks, and every four pixels are taken as a group, which is recorded as b _km , m=1, 2, 3, 4; the internal pixels of b _km are expressed as n=1,2,3,4; denoted as k represents the first large block, m represents the small block in the large block, and n represents the specific pixel; four pixels including no mark, horizontal line, diagonal line and vertical line are taken as a group, and the middle The black marked pixels are used to make the difference with their own pixels, and the black marked pixels are the embedding points of the watermark;

Step 3: Put Three pixels are rounded down to predict predicted value of pixel

Step 4: Put The actual value and predicted value of make difference and so on, the The actual value and predicted value of make difference

Step 5: Use the difference Generate a difference histogram;

Step 6: Perform translation processing on the difference histogram. For the part of the difference histogram whose pixel value is greater than 0, shift one unit to the right as a whole, and leave a blank at the position where the difference histogram pixel value is 1;

Step 7: According to the difference histogram, determine the embedding capacity, select the watermark image, and embed the watermark; specifically, the following steps are included:

Step 7.1: Start watermark embedding from the first block, and judge whether the first embedding point is a pixel that can embed a watermark; if the difference If it is 0, it means that this pixel point is an embeddable point, and the watermark is embedded; if the difference If it is not 0, the watermark will not be embedded, and continue to judge the second embedding point in the block until the four points in the 4*4 block are judged in turn, and then continue the above process for the next block, Then proceed in order;

Step 7.2: Process the watermark image, make the binary watermark image into a watermark sequence for embedding, set a watermark embedding termination value according to the size of the watermark image, this termination value is the size of the watermark image, and the image after embedding the watermark The pixel values look like this:

Among them, I(i,j)(0≤I(i,j)≤255) is the gray value of the original image, W is the watermark sequence, and I ^W (i,j) represents the pixel value of the image after embedding the watermark;

Step 8: watermark extraction and image recovery; specifically include the following steps:

Step 8.1: Extract the watermark from the first block, and judge each embedding point; if The difference is 1, indicating that the embedded watermark value is 1, and 1 is extracted; if the difference is 0, indicating that the embedded watermark value is 0, extract 0; then continue to judge the second embedded point in the block until the four points in the 4*4 block are judged in turn, and then continue to the next Carry out the above process in blocks, and then proceed in order;

Step 8.2: After extracting the watermark, restore the image, and shift the pixels with a difference greater than 1 in the difference histogram to the left to obtain the original image; the gray value of the original image is as follows:

Beneficial technical effects brought by the present invention:

The invention discloses a reversible information hiding method based on block prediction and difference histogram translation, aiming at the problem that the steps of the reversible watermarking algorithm are cumbersome and troublesome to realize, and at the same time improve the embedding capacity of the watermark, under the premise of maintaining the visual effect of the image, The algorithm steps are simplified, before embedding and extracting the watermark, there is no need for complex preprocessing, it can be realized only by dividing into blocks, and the embedding capacity of the watermark is increased in the difference histogram algorithm; compared with the traditional selection of 3*3 Blocking, using the surrounding eight pixels to predict the central pixel, can increase the number of pixels that can be embedded, thereby increasing some embedding capacity; compared with the direct use of histogram translation, it can increase the steepness of the watermark histogram The feature, highlighting the number of peak point pixels, can avoid the problem of uniform distribution of pixel values, which leads to the small capacity of the histogram translation embedded watermark; at the same time, it can improve the PSNR value and improve the visual effect of the watermarked image; the application proposes three The neighborhood block method makes block prediction according to the smoothness of the image, divides the carrier image into 4*4 blocks, and then blocks into groups of four pixels, and finally implements the prediction difference histogram translation in the block , realize watermark embedding, and then embed their own label information through reversible watermarking technology, judge whether the ciphertext has been damaged through the integrity of the watermark, and finally restore the original image without distortion; the method of the present invention can not only realize reversible watermarking, but also make the watermark invisible The performance is good, the experiment is simple and easy, and the PSNR of the image after embedding the watermark is high; the method of the invention can not only increase the embedding capacity but also ensure the accuracy of the neighborhood prediction, the image visual effect is good, and it has practical value.

Description of drawings

Figure 1 is a schematic diagram of a 4*4 block image.

Fig. 2 is a schematic diagram of a block image in the present invention.

Figure 3 is a flow chart of watermark embedding.

Figure 4 is a flow chart of watermark extraction.

Detailed ways

Below in conjunction with accompanying drawing and specific embodiment the present invention is described in further detail:

A reversible information hiding method based on block prediction and difference histogram translation, comprising the following steps:

Step 1: Set the original image as an M*N grayscale image, and divide the original image into 4*4 image blocks, denoted as B _k , where k=1,2,3...[M/4]* [N/4]; as shown in Figure 1.

Step 2: Each 4*4 image block B _k is divided into blocks, and every four pixels are taken as a group, which is recorded as b _km , m=1, 2, 3, 4; the internal pixels of b _km are expressed as n=1,2,3,4; denoted as Make In the middle of the four pixels, k represents the number of large blocks, m represents the small block in the large block, and n represents the specific pixel. The four pixels of each mark (no mark, horizontal line, oblique line, vertical line) are used as a group, and the black marked pixels in the middle are used to make a difference with their own pixels. The black marked pixels are the embedding points of the watermark, as shown in the figure 2.

Step 3: Take advantage of Three pixels are rounded down to predict predicted value of pixel

Step 4: Put The actual value and predicted value of make difference and so on, the The actual value and predicted value of make difference

Step 5: Use the difference Generate a difference histogram;

Step 6: Divide the original image into blocks and judge Whether it is a pixel that can be embedded with a watermark, if the difference is 0, it means that this pixel is an embeddable point, and a 1-bit watermark is embedded; if the difference is not 0, the watermark is not embedded, perform steps 1 and 2, and continue Judge the next pixel;

Watermark embedding, its process is shown in Figure 3,

Suppose the size of the image is M×N, and its gray value is I(i,j)(0≤I(i,j)≤255), convert the watermark image into a watermark sequence W, which is 64*64 in this application The binary watermark picture is converted into a watermark sequence of 1*4096, and the pixel value of the image after embedding the watermark is represented by I ^W (i, j);

The specific process of watermark embedding is as follows:

First, the original image is divided into blocks according to the methods shown in Figure 1 and Figure 2, and the embedding point of the original image can be known after the block is divided.

Then calculate the difference value for the divided pixel blocks, and then generate a difference value histogram from the obtained difference value. Perform translation processing on the difference histogram, and for the part of the difference histogram whose pixel value is greater than 0, shift one unit to the right as a whole, so that a blank will be left at the position where the difference histogram pixel value is 1.

Start watermark embedding from the first block, judge the embedding point, whether the difference is zero, if the difference is zero, then embed the watermark; if the difference is not zero, continue to the second embedding point in the block For judging, after the four points in the 4*4 block are judged sequentially, continue the above process for the next block, and then proceed in order. The formula looks like this:

The watermark image is processed, and the binary watermark image is made into a watermark sequence for embedding, and a watermark embedding termination value is set according to the size of the watermark image, and the termination value is the size of the watermark image.

The process of watermark extraction and image restoration is shown in Figure 4.

Firstly, the watermark is extracted, and the image embedded with the watermark is processed, and the block method is still in accordance with the block processing shown in Figure 1 and Figure 2.

still use three pixels predicted The predicted value of the pixel, and then use the predicted value and The true value of is the difference.

Then use the difference to generate a difference histogram, because the watermark is embedded when the difference is 0, so the difference is 1 to prove that the embedded watermark value is 1, and then the same as the order of embedding watermarks, from the first point The block starts watermark extraction, and judges each embedding point. If the difference is 1, the embedded watermark value is 1, and 1 is extracted; if the difference is 0, the embedded watermark value is 0, and 0 is extracted. Then continue to judge the second embedded point in the block, and after the four points in the 4*4 block are judged in turn, continue to perform the above process on the next block, and then proceed in order.

Because the pixels whose difference value histogram is greater than 0 are shifted to the right by one unit when the watermark is embedded, so before embedding the watermark, the number of pixels with a pixel value of 1 is 0, so the obtained difference is judged when extracting the watermark, if The obtained difference point is 0, indicating that the embedded watermark value is 0, and the difference value of 1 indicates that the embedded watermark value is 1. The formula looks like this:

Finally, after the watermark is extracted, the image is restored, and the pixels with a difference greater than 1 in the difference histogram are shifted one bit to the left to obtain the original image.

Since the reversible watermarking algorithm is usually more complex, the histogram of block prediction difference proposed by the present invention is relatively simple to implement for grayscale images. The embedding capacity of the watermark is relatively small, the embedding and extraction of the watermark are simple and easy to understand, and the PSNR (Peak Signal to Noise Ratio, peak signal-to-noise ratio) value is above 56. The modification of the pixel value of the image is very small, and after the watermark is proposed, the image can be restored without distortion, and the reversible watermark can be realized.

PSNR is peak signal-to-noise ratio, which is an objective standard to measure image distortion or noise level. The larger the PSNR value between two images, the smaller the image distortion. The general benchmark is 30dB, and image degradation below 30dB is more obvious. The formula looks like this:

MAX represents the maximum value of the image color, MSE is the mean square error between m×n monochrome images I and K, and the definition formula is as follows:

Use PSNR to evaluate the quality of the extracted watermark and the quality of the recovered carrier image. It can be seen from Table 1 and the experimental results that a 64*64 watermark is embedded in the grayscale image. It can be seen from PSNR that the image after embedding the watermark The overall PSNR is greater than 56, which can achieve a good visual effect, ensure the invisibility of the watermark and the effect of the image, and after the watermark is extracted, the original image and the watermark image can be restored without distortion.

Table 1 Comparison results of embedding capacity and PSNR of embedding watermark

Of course, the above descriptions are not intended to limit the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or replacements made by those skilled in the art within the scope of the present invention shall also belong to the present invention. protection scope of the invention.

Claims (1)

1. A reversible information hiding method based on block prediction and difference histogram translation, characterized in that: comprise the steps: Step 1: Set the original image as an M*N grayscale image, and divide the original image into 4*4 image blocks, denoted as B _k , where k=1, 2, 3...[M/4]* [N/4]; Step 2: Each 4*4 image block B _k is divided into blocks, and every four pixels are taken as a group, which is recorded as b _km , m=1, 2, 3, 4; the internal pixels of b _km are expressed as Referred to as k represents the first large block, m represents the small block in the large block, and n represents the specific pixel; four pixels including no mark, horizontal line, diagonal line and vertical line are taken as a group, and the middle The black marked pixels are used to make the difference with their own pixels, and the black marked pixels are the embedding points of the watermark; Step 3: Put Three pixels are rounded down to predict predicted value of pixel Step 4: Put The actual value and predicted value of make difference and so on, the The actual value and predicted value of make difference Step 5: Use the difference Generate a difference histogram; Step 6: Perform translation processing on the difference histogram. For the part of the difference histogram whose pixel value is greater than 0, shift one unit to the right as a whole, and leave a blank at the position where the difference histogram pixel value is 1; Step 7: According to the difference histogram, determine the embedding capacity, select the watermark image, and embed the watermark; specifically, the following steps are included: Step 7.1: Start watermark embedding from the first block, and judge whether the first embedding point is a pixel that can embed a watermark; if the difference If it is 0, it means that this pixel point is an embeddable point, and the watermark is embedded; if the difference If it is not 0, the watermark will not be embedded, and continue to judge the second embedding point in the block until the four points in the 4*4 block are judged in turn, and then continue the above process for the next block, Then proceed in order; Step 7.2: Process the watermark image, make the binary watermark image into a watermark sequence for embedding, set a watermark embedding termination value according to the size of the watermark image, this termination value is the size of the watermark image, and the image after embedding the watermark The pixel values look like this: Wherein, I(i, j) (0≤I(i, j)≤255) is the gray value of the original image, W is the watermark sequence, and I ^W (i, j) represents the pixel value of the image after embedding the watermark; Step 8: watermark extraction and image recovery; specifically include the following steps: Step 8.1: Extract the watermark from the first block, and judge each embedding point; if The difference is 1, indicating that the embedded watermark value is 1, and 1 is extracted; if the difference is 0, indicating that the embedded watermark value is 0, extract 0; then continue to judge the second embedded point in the block until the four points in the 4*4 block are judged in turn, and then continue to the next Carry out the above process in blocks, and then proceed in order; Step 8.2: After extracting the watermark, restore the image, and shift the pixels with a difference greater than 1 in the difference histogram to the left to obtain the original image; the gray value of the original image is as follows: