Research On Adaptive Image Watermarking Based On DCT

Digital watermarking is a process of embedding some digital information into the original data without causing perceptual distortion, and watermarking can be extracted from them. It's one of the powerful tools to protect the rights of digital works. As a rising subject, many problems remain unresolved. The main purpose of this thesis is to solve these problems, such as scaling parameter and embedded image assessment and etc. And an adaptive image watermarking algorithm is describeb here.Here we analyze some existent problems in digital watermarking, such as embedding place, scaling parameter and embedded image assessment. Then an adaptive image watermarking algorithm based on DCT is proposed to solve these problems. First significative image watermarking is pretreated. And a new universal image quality index is selected to access the embedded image distortion. The embedding process is modified continuously in order to find appropriate scaling parameters. According to the JPEG brightness table, the proportion of embedding strength of middle frequency coefficients is confirmed. And a new universal image quality index which is superior to PSNR and MSE is introduced here to analyze the embedded image distortion. If it can't be close to the expected distortion, the scaling parameter is adjusted by dichotomy so as to obtain optimal embedding effect. So watermarking are embedded into the middle frequency coefficients with different strength.Testing data indicate that local characteristic of image is better combined in our algorithm. Experimental results show that the extracted watermarking and original watermarking is resemble, so the adaptive algorithm is invisible and robust to certain common distortions.