Steganography on RGB Images Based on a “Matrix Pattern” using Random Blocks

Abstract

Abstract-In this paper, we describe a novel spatial domain method for steganography in RGB images where a secret message is embedded in the blue layer of certain blocks. In this algorithm, each block first chooses a unique t 1 xt 2 matrix of pixels as a "matrix pattern" for each keyboard character, using the bit difference of neighbourhood pixels. Next, a secret message is embedded in the remaining part of the block, those without any role in the "matrix pattern" selection procedure. In this procedure, each pattern sums up with the blue layer of the image. For increasing the security, blocks are chosen randomly using a random generator. The results show that this algorithm is highly resistant against the frequency and spatial domain attacks including RS, Sample pair, X 2 and DCT based attacks. In addition, the proposed algorithm could provide more than 84.26 times of capacity comparing with a competitive method. Moreover, the results indicated that stego-image has almost 1.73 times better transparency than the competitive algorithm.

Block Texture Pattern Detection Based on Smoothness and Complexity of Neighborhood Pixels

“…On the other hand, MP in "S_S" block texture pattern does not increase the capacity largely and so it is not very useful. This is because the same neighborhood pixels in "S_S" do not allow MP method to form different "matrix patterns" for each character [21]. In addition, "S_C" of MP method provide the best PSNR among all methods and all types of BTPs, except "S_S" of MP which cannot hide a large number of characters.…”

“…A well-known algorithm, PVD, is more efficient when it can hide data in an area within edges [20,28]. In addition, the new matrix pattern based steganography system presented earlier by the authors can produce matrix patterns faster in the blocks of image with high complexity, and it can be used for assigning more parts of the blocks of image for inserting secret message [21]. Moreover, the algorithm presented in that article calculates the difference between the neighborhoods' pixels to produce a "matrix pattern" and distribute it through the whole block.…”

“…Moreover, the algorithm presented in that article calculates the difference between the neighborhoods' pixels to produce a "matrix pattern" and distribute it through the whole block. Thus, in the presented algorithm [21], the best block texture patterns for data hiding may be those having both high smoothness and complexity, as well as fine texture in smooth area. On the other hand, LSB, the most well-known data hiding algorithm in spatial domain, can be used in different texture patterns and gets good results.…”

“…Note that, "I" refers to the cover image and "S" refers to the stego-image [21]. Finally, PSNR is calculated with (16).…”

“…The results show that this method can determine the texture blocks base on smoothness and complexity of an image with high confidence. Next, the effects of this texture classification are illustrated on three different steganography algorithms in spatial domain which are include Least Significant Bit (LSB) [19], Pixel Value Differencing (PVD) [20] and a steganography algorithm based on Matrix Pattern (MP) [21].…”