The Study Of Image Authentication Based On Content

Image authentication is an efficient mean to look after the security of image storage and transmission in the information era, and it has two aims: tamper detection and location. Traditional signature based on the text can't be used in image authentication directly. The paper mainly research on digital signature used in image authentication, and three schemes are proposed which get good results in improving probability and precision of tamper detection and resume image approximately.Scheme one is a signature based on Hash function. First, the probability of Hash Hamming weight is simulated when the inputs of Hash function change, the results indicate that the probability is 92% at least, and much higher than parity's 50%; second, use the Hamming weight as a generation function to get the signature. Theory and simulation prove that scheme is efficient to improve the precise of tamper detection and location.Scheme two and three present a new approximate image authentication method, the image is divided into regions and each region has some different 8×8 blocks, a specified fourteen DCT coefficients of blocks are RS encoded and the check bits are considered as the signature. A new location method —region location is proposed. The two schemes can not only distinguish compression from tamper, but also locate tamper and approximate error correct, and the precise is 8×8 pixel, the size of resume is 1/16 of image. The realization of the scheme is simple and robust, and the experiment results prove that it is validity.