CN108229194A - Method for Digital Image Scrambling based on multithreading model and multistage scramble - Google Patents

Abstract

The invention belongs to image encryption technical fields, are related to the digital image encryption method of multithreading model, Information hiding and multistage scramble.Make full use of Pixel-level and the advantage of bit-level scramble, different bit-level shuffle operations are placed in different threads and concurrently run, the characteristics of reducing the speed of service of bit-level scramble, and combining Pixel-level scramble, devises the digital Image Scrambling framework based on multithreading model and multistage scramble.This disorder method can be embedded in the scramble stage of any resume image.Compared to traditional disorder method, this method can significantly change the demographic information of plaintext in the scramble stage, and with the characteristics of safe, timeliness is preferable.

Description

Digital image scrambling method based on multi-thread model and multi-level scrambling

technical field

The invention belongs to the technical field of image encryption, and relates to a multi-thread model, information hiding and multi-level scrambling digital image encryption method.

Background technique

In 1997, American researcher Jiri Fridrich first proposed a scrambling diffusion framework for digital image encryption, namely the Confusion-Diffusion framework. In this framework, the digital image encryption algorithm is divided into two relatively independent stages in the process of execution, namely, the scrambling stage and the diffusion stage. In the scrambling stage, all the pixels of the image are rearranged, and the relative positions of the pixels change, but the values of the pixels do not change. In the diffusion phase, the encryption algorithm system changes the value of each pixel. In recent years, almost all digital image encryption algorithms are based on the scrambling diffusion framework.

Scrambling algorithm, as the core algorithm of digital image encryption system, has attracted the attention of many researchers in recent years and has made great progress. A series of scrambling algorithms have been proposed successively, such as row and column swapping scrambling algorithms based on sorting, scrambling algorithms based on various two-dimensional chaotic maps, and scrambling algorithms based on three-dimensional reversible maps. However, these scrambling algorithms still cannot solve the problems of insufficient scrambling, inability to change plaintext statistical information, and slow execution efficiency. In order to solve the above problems, the researchers introduced the scrambling operation into the bit-level matrix, that is, the value of the pixel point on the macro level is changed by changing the position of the bit on the micro level. However, bit-level scrambling requires more runtime due to the multiplication of scrambling planes. Pixel-level scrambling can encrypt the position matrix of pixels, while bit-level scrambling aims to change the position and value of pixels at the same time. The pixel-level scrambling algorithm can quickly perform scrambling operations, but its scrambling effect is poor, and it cannot change the statistical information of plaintext. Bit-level scrambling can solve the problem of pixel-level scrambling, but the execution process is more complicated and the operation speed is slower. After comprehensively considering the impact of different levels of scrambling on ciphertext, the present invention proposes a scrambling algorithm that comprehensively considers the characteristics of pixel level and bit level.

The popularity of multi-core CPU provides a hardware basis for parallel computing. The present invention fully considers the characteristics of the underlying hardware platform when designing, and designs a digital image scrambling algorithm based on a multi-thread model and multi-level scrambling. This algorithm can run on different multi-thread platforms, such as Windows SDK thread library and POSIX Thread thread library.

Contents of the invention

In order to make full use of the advantages of pixel-level and bit-level scrambling, put different bit-level scrambling operations into different threads to run concurrently, reduce the running speed of bit-level scrambling, and combine the characteristics of pixel-level scrambling, design a Digital image scrambling architecture for multi-threaded models and multi-stage scrambling.

Technical scheme of the present invention is:

A digital image scrambling method based on a multi-thread model and multi-stage scrambling, comprising the following steps:

Step 1: Separate the color channels of a three-channel color image to obtain three single-channel images of B, G, and R, and then divide the three single-channel images into bit layers, and divide each image into 8 Bit plane, a total of 24 binary images are obtained;

Step 2: Assign a thread to each bit plane, so that each bit plane independently executes the bit-level scrambling operation on its own thread, and multiple threads execute concurrently; after the scrambling process is completed, 24 scrambled Bit-level scrambling graph, the first level of scrambling ends;

Step 3, the second level of scrambling is completed at the pixel level; the 24 scrambled bit-level planes are restored to single-channel images of B, G, and R components, and then pixel-level scrambling is performed on the three single-channel images. scrambling, the secondary scrambling ends at this time;

Step 4, the three single-channel images after scrambling are merged into a three-channel color image, and the final scrambling image is obtained at this time.

Further, in the above step 2, according to the characteristics of the target image, the bit planes of each single-channel image are divided into high and low groups; Iteratively derived; for low-bit planes, only 2D random map scrambling is performed.

The above-mentioned two-dimensional chaotic system adopts the cat map two-dimensional scrambling method; the one-dimensional chaotic system adopts the logistic map one-dimensional scrambling method.

The above method also includes a decryption process, which is the inverse process of the scrambling encryption process: first, the color channel separation is performed on the ciphertext image to obtain three single-channel images of B, G, and R, and then the reverse operation of pixel-level scrambling is performed. The second-level scrambling is restored; then, the bit planes of B, G, and R are divided to obtain 24 bit planes; a thread is assigned to each bit plane, and each bit plane independently performs the scrambling restoration operation on its own thread , multiple threads execute concurrently; the 24 bit planes obtained at this time are the original binary image, which are combined, and the obtained 3-channel image is the original image after decryption.

The beneficial effect of the present invention is that, by comprehensively considering the respective characteristics of pixel-level scrambling and bit-level scrambling, the method designs a multi-level scrambling method based on multithreading. This scrambling method can be embedded in the scrambling stage of any image encryption algorithm. Compared with the traditional scrambling method, this method can significantly change the statistical information of the plaintext in the scrambling stage, and has the characteristics of high security and good timeliness.

Description of drawings

Figure 1 shows the 8 bit planes of the blue channel image, (a) ~ (h) from high to low.

Figure 2 is the 8 bit planes of the green channel image, (a) ~ (h) from high to low.

Figure 3 is the 8 bit planes of the red channel image, (a) ~ (h) from high to low.

Fig. 4 is a flow chart of digital image encryption based on multi-thread model and multi-level scrambling.

Fig. 5 is a flow chart of digital image decryption based on multi-thread model and multi-level scrambling restoration.

Figure 6 is a histogram of scrambled images; (a) B channel, (b) G channel, (c) R channel.

Detailed ways

1. Overall implementation principles and detailed steps

This method is based on a multi-threaded model. For 24 bit planes divided by a three-channel color image, a thread is assigned to each bit plane, and each plane performs scrambling operations independently on its own thread, so that these threads are concurrent Execute and process images in parallel to improve the efficiency of the algorithm.

Step 1: read a 3-channel color image, perform color channel separation operation on the image, obtain 3 single-channel images B, G, R, and then divide these 3 single-channel images into bit planes to obtain 24 Bit planes (i.e. 24 binary images).

Step 2: Allocate a thread for the upper 4-bit plane of each single-channel image, each plane independently executes the scrambling operation on its own thread, and multiple threads execute concurrently. The specific operation is as follows:

(1) Generate 12 threads, allocate a thread for the high 4-bit plane of each single-channel image, 12 threads execute concurrently, and scramble the image at the same time;

(2) In each thread function, first perform two-dimensional random map scrambling on the bit plane, here is the cat map two-dimensional scrambling method: map the pixel values of the binary image to a two-dimensional array one by one In , two parameters p and q (p, q<image_size) are used to determine the target position of the mapping, and the pixel is mapped from the original position to the target position, and the effect of scrambling is achieved by changing the relative position of the pixel. Obtain now the two-dimensional array catmapArr _m after scrambling (m represents different bit planes), each point in the array corresponds to a pixel value in the binary image;

(3) Use one-dimensional random map scrambling to generate an array that is as large as the current bit plane. The values in the array are random in size. Here, the logistic map one-dimensional scrambling is used: the value of x _i is recursively generated, and the initial value x ₀ is the secret Key value; then convert the obtained one-dimensional random array into the corresponding two-dimensional array logisticArr _m (equal to catmapArr _m );

(4). XOR the pixel values at the corresponding positions in the two two-dimensional arrays of logisticArr _m and catmapArr _m , and save the result in the two-dimensional array afterConfsuion _m . The values in this array are the pixel values of the scrambled image .

afterConfsuion _m (i, j) = logisticArr _m (i, j) ⊕ catmapArr _m (i, j);

(5) restore the pixel value in the two-dimensional array afterConfsuion _m in each thread function to the corresponding position of the binary image, and complete the scrambling of the high bit plane at this time;

(6) When all the 12 parallel threads are executed, close the thread handle object.

Step 3: Allocate a thread for the low 4-bit plane of each single-channel image, each plane independently executes the scrambling operation on its own thread, and multiple threads execute concurrently. The specific operation is as follows:

(1) Generate 12 threads, allocate a thread for the low 4-bit plane of each single-channel image, 12 threads execute concurrently, and scramble the image at the same time;

(2) In each thread function, carry out two-dimensional random map scrambling to this bit plane, what adopt here is cat map two-dimensional random map scrambling method, obtain the two-dimensional array low_catmapArr _m after scrambling;

(3) Restore the pixel value in the two-dimensional array _{low_catmapArrm} in each thread function to the corresponding position of the binary image, and complete the scrambling of the low bit plane at this moment;

(4) after all executions of 12 parallel threads are finished, close the thread handle object;

At this point, the first level of scrambling ends, and 24 scrambled bit planes are obtained.

Step 4: Merge the 24 scrambled bit planes into 3 single-channel images. At this time, based on the three pixel-level component planes, perform secondary scrambling, using a two-dimensional scrambling sorting-based permutation method.

Step 5: Merge the 3 single-channel images after the secondary scrambling into a 3-channel image, and write it into a file.

2. Take the correlation coefficient as an example to analyze the algorithm

The scrambled and encrypted image is divided into three single-channel images B, G, and R, and the horizontal correlation, vertical correlation and diagonal correlation analysis are performed on these three single-channel images respectively.

Table 1 Correlation coefficient of adjacent pixels in the image after scrambling

As can be seen from the above test results, the correlation coefficients of all aspects of the image have declined, even approaching zero, and thus it can be concluded that the multi-stage scrambling method of the present invention has achieved an ideal effect. Scrambling makes the logical correlation of the image completely disrupted.

3. Take the histogram as an example to analyze the algorithm

Divide the scrambled and encrypted image into three single-channel images B, G, and R, draw and analyze histograms for these three single-channel images

From the analysis of the histogram in Figure 6, the number of each pixel value is roughly on the same horizontal line, and the scrambling effect is ideal.

4. Taking information entropy as an example to analyze the algorithm

Divide the scrambled encrypted image into three single-channel images B, G, and R, and calculate the information entropy of these three single-channel images respectively

Table 2 Information entropy analysis of scrambled image

B channel G channel R channel information entropy 7.957353520517321 7.987688899393857 7.965938174720284

Information entropy refers to the degree of information confusion; observing the above test data, the information entropy of the image after scrambling and encryption increases significantly, and the entropy value is close to the ideal value of 8, and the effect of scrambling is ideal.

5. Take the throughput of the scrambling algorithm as an example to analyze the algorithm

The throughput according to the program is: Throughput=142.513K/s.

To sum up, the original scrambling algorithm only performs a first-level scrambling for the entire image. However, the present invention raises it from one-level scrambling to multi-level scrambling, which greatly increases the security of hidden information; and adopts a multi-thread model to realize, and improves the execution efficiency of the algorithm. Therefore, the method can be better used in application processes such as information hiding and encrypted transmission in the field of information security.

Claims (4)

1. the Method for Digital Image Scrambling based on multithreading model and multistage scramble, which is characterized in that include the following steps：

Step 1 carries out a Three Channel Color image separation of Color Channel, obtains tri- single channel images of B, G, R, then Respectively these three single channel images are carried out with the division of bit-level, every width figure is divided into 8 bit-planes, is obtained 24 two It is worth image；

Step 2 distributes a thread for each bit-planes, each bit-planes is enabled to be independently executed on respective thread Bit-level shuffle operation, multiple threads concurrently perform；After the completion of the scrambling process, the bit-level scramble after 24 scrambles is obtained Figure, first order scramble terminate；

Step 3, second level scramble are completed in pixel scale；Bit-level plane after 24 scrambles is reduced to as B, G, R tri- The single channel image of component, and then Pixel-level scramble is carried out to three single channel images, Second-scrambling terminates at this time；

Three single channel images after scramble are merged into a Three Channel Color image, obtain final put at this time by step 4 Random image.

2. the Method for Digital Image Scrambling according to claim 1 based on multithreading model and multistage scramble, feature exist In in step 2, according to target image characteristics, the bit-planes of each single channel image are divided into two groups of height；For high ratio Special plane, the mapping ruler in scramble are obtained by Two Dimensional Chaotic Systems and one-dimensional chaos system common iterations；For low bit Plane only carries out two-dimensional random mapping scramble.

3. the Method for Digital Image Scrambling according to claim 2 based on multithreading model and multistage scramble, feature exist In the Two Dimensional Chaotic Systems use cat map two dimension disorder methods；One-dimensional chaos system is one-dimensional using logistic map Disorder method.

4. the Method for Digital Image Scrambling based on multithreading model and multistage scramble according to claims 1 or 2 or 3, It is characterized in that, further includes decrypting process, be the inverse process of scrambling encryption process：Color Channel point is carried out to ciphertext image first From obtaining tri- single channel images of B, G, R, then carry out the contrary operation of Pixel-level scramble, reduce Second-scrambling at this time；And then The division of bit-planes is carried out to B, G, R, obtains 24 bit-planes；A thread is distributed for each bit-planes, each Bit-planes independently perform scramble restoring operation on respective thread, and multiple threads concurrently perform；24 ratios obtained at this time Special plane is original bianry image, is merged, and 3 obtained channel images are the original image after decrypting.