A study on preprocessing method for steganalysis feature extraction

The application of image steganography has greatly satisfied the needs of individuals and public security communication. Although steganography technology provides a convenient way to protect personal information exchange, it also becomes an effective tool for criminals. Steganalysis studying is very necessary. The position and capacity of secret information in stego images depend on the sharpness of the region in the adaptive steganography algorithms in spatial domain. According to this feature we propose a region segmentation pre-processing method in this paper, which is based on absolute difference of gray intensity. In order to improve the features discrimination, the entire image is divided into three different regions roughly, and then features are extracted. Those features have a higher correlation according to region characteristics. This proposed method offers a significant improvement in detection accuracy when compared to Region Segmentation method.