CN104284190B - Compressed image steganography encoding method based on AMBTC high and low average optimization - Google Patents

Abstract

The invention discloses a compressed image steganography encoding method based on AMBTC high and low average optimization. According to the compressed image steganography encoding method disclosed by the invention, after the smoothness of an image block is evaluated, data are embedded in smooth blocks one by one to obtain a stego image, wherein the data embedding comprises two steps: replacing a binary bitmap by the data, and optimizing high and low averages according to the embedded data and an original bitmap. With the adoption of the compressed image steganography encoding method, the computing is simple, the method is easy to realize, and a receiving side only needs to directly take out a bitmap of a receiving side, namely the embedded data, by discriminating the smooth blocks according to a secrete key; in addition, requirements on computing resources are low, and the use is convenient, so that the compressed image steganography encoding method has a good application prospect in the real-time field, mobile terminals with limited computing resources, and the like.

Description

One kind is based on the optimized compression image latent writing coded method of AMBTC height averages

Technical field

The present invention relates to information security, secret communication and multimedia application field, and in particular to one kind is based on AMBTC height The optimized compression image latent writing coded method of average.

Background technology

Present information steganography is the important directions of digital information epoch information security field, and steganography technology is sharp With the coding redundancy and structural redundancy of Digital Media itself, and the insensitivity of human perception organ, by steganography in load In the middle of volume data, carrier signal can be text, image, video, audio frequency etc..And steganography can further with cryptography, letter The multidisciplinary multi-field combination such as breath treatment technology, network technology and physiology and psychology, in industrial quarters, business circles and army's event horizon tool Have broad application prospects, such as covert communications, information labeling and distorts certification etc. at digital finger-print.

For carrier, natural image is high due to redundancies such as itself structure, codings, it is common be easy to get grade easy to use into It is using most steganography carriers.Steganography with digital picture as carrier also most worthy.Wherein image latent writing again can be The different aspects such as image space domain, ciphertext domain and compression domain are carried out, and the steganography with the compressed encoding of image as carrier is as compressed Image latent writing.Due to image after compressed encoding carrier data redundancy reduce, therefore compress image in steganographic capacity and Steganography quality is often relatively weak.

AMBTC (Absolute Moment Block Truncation Coding) codings are a kind of simple and effective fast Fast Image Lossy Compression technology, compares other compressed encodings such as JPEG, and fast with coding rate, algorithm complex is low, takes money The features such as source is few, is quite paid attention in real time imaging transmission field.Such as CellB video formats, the Xmovie and DEC of SUN companies SMP of company etc..AMBTC encoders need to compress first image division into nonoverlapping piece of size, for each sub-block is calculated Pixel value is 1 more than or equal to the pixel logo of average by pixel average, binary bitmap B, otherwise for 0.In this sub-block, calculate 1 pixel average is designated, high average h is designated as；Another kind of pixel average is calculated, low average l is designated as, (l, h, B) ternary Group is the coding of each sub-block.During decoding AMBTC compression images, each sub-block compressed code (l, h, B) is read from code stream first, Then sweep bitmap B, on the contrary it is worth and is redeveloped into high average h for 1 and is then redeveloped into low average l.

Steganography at present with AMBTC image compression encodings as carrier is few, and in terms of steganography is damaged, main thought is such as Under.Chuang in 2006 et al. is directly embedded data into flat using the high average of the smooth block property little with low average difference In the binary bitmap of slide block, but hidden image Quality Down is serious；2008, Hong et al. was by big between height average Little relation pair answers bitmap upset to realize reversible steganography, does not affect picture quality but each sub-block can only be embedded in a binary data And it is unavailable to change block when height average is equal；Chen in 2010 et al. is further improved on the basis of Hong, when height The bitmap block is enabled when being worth equal for embedding data to improve algorithm capacity, but whole volume is still less；Hong in 2011 Et al. further improve bitmap on the basis of Chuang2006 again, all this kind innovative point is main to improve picture quality ... Concentrate on the bitmap of upset height average and bit plane or replacement or modification smooth block.Replace smooth using data to be embedded Bitmap block is to ensure the good thinking of BTC domains steganographic capacity, but while picture quality also declines seriously, it is ensured that the premise of capacity Under, how to be a new thinking by trying to achieve the height average optimal solution with embedded Data Matching to improve picture quality.

The content of the invention

Goal of the invention：It is an object of the invention to solve the deficiencies in the prior art, the present invention is in smooth bitmap block Under replacing thinking, start with from impact of the analysis height average to picture quality, height is tried to achieve by proving minimum mean square error Average optimal correction value, it is final to provide a kind of based on the optimized compression image latent writing coded method of AMBTC height averages.

Technical scheme：One kind of the present invention compresses image latent writing coded method based on AMBTC height averages are optimized, according to It is secondary to extract two steps with data including data are embedded, wherein, data are embedded to be included：

(1) image block smoothness is assessed；

(2) embedding data；

(3) all image blocks are processed and obtains carrying close image.

Further, the concrete grammar of the step (1) is：

Image is compressed to AMBTCWherein, (l_i,h_i,B_i) it is AMBTC ternarys Group, parameter l_i、h_iAnd B_iLow average, high average and the bitmap of each image block are represented respectively, and n and m represents respectively the block of image block Number and block size, when parameter meets formula (1), decision block (l_i,h_i,B_i) it is smooth block, T is threshold value, and using T as shared Key；

h_i-l_i≤T (1)。

Further, in step (2), m positions data to be embedded are takenAnd embedded smooth block (l_i, h_i,B_i) in, specific Embedded step is as follows：

(2.1) bitmap is countedWith embedded dataThe change feelings of correspondence position Condition k_pqExpression meets conditionK number, it is clear that

(2.2) according to formula (2), the height average optimal value (l ' after adjustment is calculated_i,h′_i),

In formula (2), work as k₀₀+k₁₀When=0, l '_i=h '_i；Work as k₀₁+k₁₁When=0, h '_i=l '_i；

(2.3) with the tlv triple (l ' after adjustment_i,h′_i,D_j) replace former tlv triple (l_i,h_i,B_i), that is, complete m bit datasIt is embedded.

Further, the concrete grammar of the step (3) is, completes when all image blocks are processed according to step (2) Afterwards, obtain and carry close image

Further, the concrete grammar of the data extraction is：

Carry close image-receptive root and close AMBTC compressions image is carried to step (3) gained according to formula (3) Detected using shared key T block-by-block, the decision block (l ' when block parameter meets formula (3)_i,h′_i,B′_i) to carry close piece, otherwise lose Abandon,

h′_i-l′_i≤T (3)；

To carrying close piece of (l '_i,h′_i,B′_i), extracting directly its m positions data bitmapAs carrying close piece Embedding data.

Further, parameter k in the step (2.1)_ijCalculating process be：

Wherein, operator | | for the base of calculations incorporated.

Further, in the step (2.2), by height average optimal value (l '_i,h′_i), obtain image block (l_i,h_i, B_i) and (l '_i,h′_i,D_j) between minimum error MSE, i.e.,：

MSE=(k₀₀(l′_i-l_i)²+k₀₁(h′_i-l_i)²+k₁₀(l′_i-h_i)²+k₁₁(h′_i-h_i)²)/m (5)。

Obtained after MSE, by MSE to l ' by formula (5)_i, h '_iSingle order second dervative is sought respectively：

Due toTherefore work asTime error MSE takes minima.

Solution Simultaneous Equations (11) can obtain formula (2).

The calculating process of above-mentioned formula (7)～formula (11), be：Height average optimal value when making MSE minimum is asked Solution preocess.Image block (l can be obtained by height average optimal value_i,h_i,B_i) and (l '_i,h′_i,D_j) between minimum error.

Further, in the step (2.2), to smooth block (l_i,h_i,B_i) formula (2) is utilized by height average (l_i,h_i) It is adjusted to (l '_i,h′_i) when, meet following condition：

Its proof procedure is as follows：

Beneficial effect：The present invention is estimated first to the smoothness height value difference of AMBTC compressed picture blocks, works as height Low equal value difference then assert that it is " smooth block " less than the threshold value (key) for setting；Then embedding data is implemented to the bitmap of smooth block Replace embedded；Finally according to embedded data point reuse height average picture quality is optimal.

The present invention calculates simple, facilitates implementation, and recipient only needs to directly take out its bitmap according to key examination smooth block As embedded data；It is easy to use and the present invention is little to computational resource requirements, it is limited in real-time domain and computing resource Mobile terminal etc. has good application prospect.

Description of the drawings

Fig. 1 is the embedded flow chart of data in the present invention；

Fig. 2 is the flow chart that data are extracted in the present invention；

Fig. 3 is the visual quality comparison diagram of the fully loaded rear decompressed images of Lena in the present invention；

Fig. 4 is the effect comparison diagram of the present invention and prior art；

Fig. 5 is the schematic diagram of each step of embodiment：

Wherein, Fig. 3 a are original image, and Fig. 3 b are fully loaded rear decompressed image；Fig. 5 a are the gray-scale maps of the 8 × 8 of embodiment Picture；Fig. 5 b are the AMBTC compressed encoding schematic diagrams of embodiment；Fig. 5 c are the image schematic diagram of Fig. 5 b decompressions；Fig. 5 d are enforcement The schematic diagram of the data to be embedded of example；The close compressed code figure of load when Fig. 5 e are T=1 in embodiment；Fig. 5 f are close for the load of embodiment Decompression figure.

Specific embodiment

Technical solution of the present invention is described in detail in conjunction with the accompanying drawings and embodiments below.

Embodiment 1：

As depicted in figs. 1 and 2, the present embodiment is comprised the following steps that：

Data it is embedded：

AMBTC compressed encodings are carried out to the gray level image of 8 × 8 as shown in Figure 5 a, as a result as shown in Figure 5 b, four is obtained Image block：(161,162, B₁), (157,162, B₂), (157,162, B₃) and (158,161, B₄), given threshold T=1 works as block Parameter meets h_i-l_iDuring≤T, can determine that image block is smooth block, it can be seen that, in the present embodiment, block (161,162, B₁) it is flat Slide block, and the decompressed image of Fig. 5 b is as shown in Figure 5 c.

16 data to be embedded as fig 5d are embedded into into smooth block (161,162, B₁) in, comprise the following steps that：

Bitmap is counted according to formula (4)To embedding data

Change such as table 1, wherein

Specific embodiment k of table 1_ijStatistics

The height average optimal value (l ' after adjustment is calculated according to formula (13)_i,h′_i)=(162,162).

Then, with the tlv triple (162,162, D after adjustment₁) replace former tlv triple (161,162, B₁), now,

Complete m=16 bit data D₁It is embedded, obtain the close compression figure of load as depicted in fig. 5e, the decompression figure of Fig. 5 e As shown in figure 5f.

The extraction of data：

Recipient compresses imagery exploitation shared key T=1 block-by-block according to formula (3) to the close AMBTC of load as depicted in fig. 5e Detection, judges image block as close piece, in the present embodiment, block 1 (162,162, D is carried when block parameter meets formula (3)₁) it is to carry Close piece, block 2, block 3 and block 4 carry close piece for non-.

To carrying close piece (162,162, D₁), extracting directly its data bitmap D₁As embedding data.

In order to detect the effect of the present embodiment, tested and analyzed in terms of steganographic capacity and picture quality two.With 512 As a example by the working standard test image of × 512 pixels, work as T=15, during m=16, the steganographic capacity and correspondence PSNR of different images Value is as shown in table 2.Wherein Bits is embedded data length, and unit is bit, and PSNR1 is to carry close compression image decompression and original Y-PSNR between figure, PSNR2 is to carry close compression image decompression to carry the peak value letter that close AMBTC encodes direct decompressed image with non- Make an uproar ratio.Y-PSNR value is that a kind of of camouflage test visual quality evaluation index quantifies reference, and research shows when two width gray scales Human eye can not be identified the difference of the two when Y-PSNR between image is more than 30.Fig. 3 a and Fig. 3 b be original image with it is fully loaded Afterwards the visual quality of decompressed image (T=4, m=16) compares.

The standard testing image steganographic capacity of table 2 and correspondence image quality T=15, m=16

For vivider statement experimental result and the beneficial effect of innovation and creation, we by taking Lena as an example, by institute of the present invention Extracting method is compared with the method for Hong and Chuang, as shown in figure 4, abscissa is threshold value T, vertical coordinate is PSNR, that is, lead to Cross Fig. 4 and can be seen that hidden image quality that under equal load (threshold value is identical) present invention produces and significantly improve.

Additionally, the safety in order to strengthen embedded data, can also use AES before steganography to embedded data It is encrypted, pseudo-random permutation image AMBTC compression code blocks is produced using key (seed) and is ranked up, different keys is produced Different image AMBTC compresses the arrangement of code block.This is a huge information space, and the assaulter for not knowing about key is difficult brokenly Translate.

Claims (7)

1. a kind of based on the optimized compression image latent writing coded method of AMBTC height averages, it is characterised in that：Include number successively Two steps are extracted according to embedded and data, wherein, data are embedded to be included：

(1) image block smoothness is assessed：

Image is compressed to AMBTCWherein, (l_i,h_i,B_i) it is AMBTC image block ternarys Group, parameter i represents the numbering of image block, parameter l_i、h_iAnd B_iLow average, high average and the bitmap of each image block are represented respectively, N and m represent respectively the block number and block size of image block, when parameter meets formula (1), decision block (l_i,h_i,B_i) it is smooth block, T is threshold value, and using T as shared key；

h_i-l_i≤T (1)；

(2) embedding data；

(3) all image blocks are processed and obtains carrying close image.

2. according to claim 1 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：In step (2), m positions data to be embedded are takenAnd embedded smooth block (l_i,h_i,B_i) in, specifically Embedded step it is as follows：

(2.1) bitmap is countedWith embedded dataSituation of change k of correspondence position_pq Expression meets conditionK number, it is clear that

k_pq∈[0,m]&∑k_pq=m, p, q ∈ { 0,1 }；

(2.2) according to formula (2), the height average optimal value after adjustment is calculated

$\left\{\begin{array}{c}{l}_i^{\′}=(l_i\×k_{00}+h_i\×k_{10})/(k_{00}+k_{10})\\ h_i^{\′}=(l_i\×k_{01}+h_i\×k_{11})/(k_{01}+k_{11})\end{array}\right.---\left(2\right);$

In formula (2), work as k₀₀+k₁₀When=0, l '_i=h '_i；Work as k₀₁+k₁₁When=0, h '_i=l '_i；

(2.3) with the tlv triple (l ' after adjustment_i,h′_i,D_j) replace former tlv triple (l_i,h_i,B_i), that is, complete m bit datasIt is embedded.

3. according to claim 1 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：The concrete grammar of the step (3) is, according to step (2) after the completion of all image blocks are processed, obtains and carries close figure PictureWherein l '_i,h′_i,B′_iHigh and low average and bitmap after steganography are represented respectively.

4. according to claim 1 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：The concrete grammar that the data are extracted is：

Carry close image-receptive root and close AMBTC compressions image is carried to step (3) gained according to formula (3)Utilize The detection of shared key T block-by-block, the decision block (l ' when block parameter meets formula (3)_i,h′_i,B′_i) to carry close piece, otherwise abandon,

h′_i-l′_i≤T (3)；

To carrying close piece of (l '_i,h′_i,B′_i), extracting directly its m positions data bitmapAs close piece be embedded in of load Data；

Wherein, l_i、h_iAnd B_iLow average, high average and the bitmap of each image block, l ' are represented respectively_i,h′_i,B′_iRepresent respectively hidden High and low average and bitmap after writing.

5. according to claim 2 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：Parameter k in the step (2.1)_ijCalculating process be：

Wherein, operator | | for the base of calculations incorporated.

6. according to claim 2 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：In the step (2.2), by height average optimal value (l '_i,h′_i), obtain image block (l_i,h_i,B_i) and (l '_i,h ′_i,D_j) between error MSE minimize, i.e.,：

MSE=(k₀₀(l′_i-l_i)²+k₀₁(h′_i-l_i)²+k₁₀(l′_i-h_i)²+k₁₁(h′_i-h_i)²)/m (5)。

7. according to claim 2 based on the optimized compression image latent writing coded method of AMBTC height averages, its feature It is：In the step (2.2), to smooth block (l_i,h_i,B_i) formula (2) is utilized by height average (l_i,h_i) it is adjusted to (l '_i, h′_i) when, meet following condition：

$h_i-l_i\≤T\⇒h_i^{\′}-l_i^{\′}\≤T---\left(6\right).$