A novel steganographic method based on edge detection and adaptive multiple bits substitution

Abstract

The present paper proposes a novel method for grayscale image steganography based on edge detection and adaptive multiple bits substitution. Compared with the smooth regions, the pixels located in the edge regions usually present more random characteristics. In proposed method, the gradient magnitude of the pixels of the cover image is computed using the Sobel operators. As a result, all edges of the cover image, both horizontal and vertical, are fully detected. According to the length of secret data, the sharper edges are adaptively preserved and the weaker edges are suppressed. Therefore, the sharper edges will be used in advance of the weaker edges and the smooth regions for data embedding. Next, the data embedding route is determined using a pseudorandom number generator (PRNG) and multiple bits of secret data are adaptively embedded into k-LSBs of the pixels in the route. The value k depends on the gradient magnitude of each pixel. The larger the gradient magnitude, the larger the value k. The experimental results evaluated on 8000 natural images with RS steganalytic algorithm show that our proposed method can significantly enhance the security and can increase the embedding capacity.