A High Quality and Capacity Steganographic Method by Pixel-value Differencing

Abstract

Image steganography is a secret communication technique that is different from encryption/decryption methods from cryptology. In 2003, Wu and Tsai proposed a simple and effective image steganographic technique, called the pixel-value differencing (PVD) technique. Its main concept and technic is to decide the amount of information bits embedded in two neighboring pixels based on the difference of their pixel values. In this way, regions with higher smoothness will embed less information and those with lower smoothness will have the opposite, therefore reducing the degree of distortion of the image and achieving imperceptibility. In addition, experimental results show that their method is capable of resisting 2 Fridrich et al.’s RS steganalysis. Because of the simplicity and efficiency of the pixel-value differencing technique, Wang, Wu, Tsai, and Hwang proposed an improved method with lower pixel value modification using modulus computation in 2008, which further increased the image quality of the steganography utilizing the pixel-value differencing technique. In this paper, we refer to Wang, Wu, Tsai, and Hwang’s method and propose a new pixel-value differencing technique implementing the concept of using “weights”, followed by carrying out a series of experiments. The results prove that our proposed method has higher stego-image quality than Wang, Wu, Tsai, and Hwang’s method when embedding the same secret message, implying that we have higher PSNR value. Especially under the requirement of high embedding capacity, our method still maintains a safe PSNR (Peak Signal to Noise Ratio) value where the PSNR value of Wang, Wu, Tsai, and Hwang’s method drops too low and fails to satisfy imperceptibility. Therefore, compared with previously proposed embedding techniques that base on pixel-value differencing, our method provides greater suitability for practical applications.