Steganalysis For Digital Image Steganography Based On LSB

With the rapid development of the digital multimedia and the Internet, steganography is widely used on the Internet. Military, government and financial organization which concern about the national economy and the people's livelihood are using this technology to transmit secret files via the Internet. But, steganography can also be easily used by malicious individuals or groups for the evil purpose, such as spreading the illegal information. Steganalysis is the counter-technology of steganography. Its purpose is to detect the existence of secret information, then destroy the invisible communication. Steganalysis is the key technology to solve the lawless use of steganography, and an important means to measure the security of steganography.This paper mainly researches on blind steganalysis for digital image steganography based on LSB. According to the difference of statistical properties between the normal images and stego-images, some steganalytic algorithms are proposed.This paper includes the following contents:(1) Concluded the research results of steganalysis, and further studied several classic steganalysis algorithms, including RS, DIH, RQP and GPC etc. Then statistical features of these algorithms were analyzed, and these algorithms' detection processes were described in detail.(2) Further researched the changes'characteristic of adjacent pixels after embedding the secret information, and a new steganalytic algorithm using correlation of adjacent pixels is proposed. The new algorithm just considers the adjacent pixel pair in which one is odd and the other is even. The algorithm only need to count the number of the adjacent pixel pair in which the odd one is greater than the even one and the number of the adjacent pixel pair in which the even one is greater than the odd one, respectively. It can be simply applied and its statistics is small. Experimental comparison shows that the new steganalysis algorithm has a strong detection capability.(3) Further researched the changes'characteristic of the sum of the 3 color components in a pixel after embedding the secret information, and a new steganalytic approach for 24-bit color images is proposed, which is based on the the frequency histogram of the sum of the 3 color components in a pixel. A classifier was constructed to identify the stego-image from the normal images, which was based on the characteristic that the frequency of the three color components'sum of adjacent pixels is closer. The new algorithm has a better performance than the RQP steganalysis method in practicability and in computation complexity.