An Adaptive Matrix Embedding for Image Steganography

In this paper, an adaptive matrix embedding method is proposed for grayscale images. The 2×4 pixel block is taken as a cover unit. The local correlation of the block is used to determine the message bit amount and choose the corresponding matrix embedding strategy. The method can embed less in image parts with high local correlation and more in image parts with low local correlation. The experimental results indicate that the new method improves the PSNR value greatly. And the new method is secure against the traditional steganalytic method.