An Effective Digital Watermarking Algorithm for Binary Text Image

摘要

Aiming at the binary text image's characteristics of simple pixel, complex texture and bad immunity of information concealment, a digital watermarking embedment location choosing method has been put forward based upon compatible roughness set. The method divides binary text image into different equivalent classes. Equivalent classes are further divided into different subclasses according to each pixel's degree and texture changes between blocks. Through properties' combination, the embedment block and location which are fit for watermarking are found out. At last, different binary text images are chosen for emulation experiment. After being embedded, the image is compressed in JPIG-2. Gaussian noise, salt & pepper noise are added and cutting is employed to imitate the actual environment in which images may suffer from various attacks and interferences. The result shows that the detector has a sound testing effect under various conditions.