A Blind Watermarking Scheme for Binary Image Authentication

A blind fragile watermarking scheme for binary image authentication is proposed. In authentication watermarking a hidden data is inserted into an image in order to detect any alterations. In the proposed method, the host binary image is first divided into non-overlapping 2 2 sub-blocks. Then, for the last position in each of the blocks, the probability of bit occurrence, conditional to different patterns of its 4-neighborhood, is calculated. The last pixel is predicted using the value of its neighbors. Bit embedding is done by modifying the predicted value instead of the real value. The proposed method is simple, fast, blind, and capable of detecting modifications. Several experiments using text binary images are presented which confirm the effectiveness of the algorithm.