Steganalysis Of LSB Matching Based On Spatial Correlation

The main research of steganography is to embed secret information into common cover media, without causing other people's suspicion. The purpose of steganography is to protect the secret information in covers by ensuring the high performance of concealment of the process. There are many kinds of cover medias used in steganography that can be downloaded through internet., such as LSB replacement in spatial domain covered by BMP images, EzStego that covered by patlette images(GIF images) and F5, Outguess covered by JPEG images in frequency domain. Steganalysis is to detect the secret information hidden in digital medias. As information security is increasingly concerned, Steganalysis technology has become one of main researchers.Our main research is LSB matching steganalysis in BMP images. LSB matching steganography is a spatial domain method with good performance of concealment. It improves the statistical asymmetrical of LSB replacement. LSB matching not only ensures visually undetectable of stego images, but also keeps the statistical features of cover images well, which makes the steganalysis harder.Firstly, we introduces the embedding and extracting process of LSB matching and analyze the statistical features of this method. Then we detailedly described some tipical steganalysis method of LSB matching, including HCF COM method, WAM algorithm in wavelet domain and ALE method.With the additive noise model, we firstly proposed a LSB matching steganalysis based on the high frequency component of images, which is to improve the performance of HCF COM methods. Then we proposed another steganalysis algorithm based on correlation features in spatial domain of images. First, the correlation of the image of which some bit planes were set zeros were calculated. Features were constructed by combining the methods of estimating cover images. At last, we compared all the algorithms by designing expriments.