The Research Of Steganography And Steganalysis Based On Visual Characteristics

With the development of digital technology, digital images become the most important vehicle of the digital steganography. Invisibility and embedding capacity are the most important concerns. The general embedding algorithm may decrease the invisibility because of the elevation of embedding capacity, while visual characteristic-based algorithms improve the invisibility and embedding capacity through concealing the confidential information in the visual areas which are not sensitive. Furthermore, it occupies an important position in the areas of digital steganography. This paper focus on the analysis and realization of three typical algorithms (including BPCS,PVD,Improved-Difference), subsequently a new detection algorithm aiming at BPCS is proposed based on the theory of run-length and support vector machine.This paper is organized as follows:(1) Four kinds of digital image formats will be analyzed, including BMP, JPEG, GIF, TIFF.Then a unified interface program is provided to support them.(2) Three typical algorithms are introduced in detail, including BPCS,PVD,Improved-Difference, which can be accomplished through unified interface program in VC++ 6.0 environment. Therefore, the achievement courses of algorithms are extremely simplified.(3) A new detection algorithm based on variety of run-length is proposed. With the number counting of the run-length in four, five, six and seven extents above the lowest three bit-planes of images, the second-class Support Vector Machines is used to distinguish the original image with the hidden one, afterwards the amount of secret information can be estimated by the use of the support vector regression. The new algorithm is compared with the one based on statistical histogram, the excellence performance of which is proved in theory and practice.