Robust Watermarking Adaptive Re-embedded

The rapid growth of the IT(Imformation Technology) makes electronic media communication easier and more extensive than before. While this enables the illegal creation and dissemination of products quickly via electronic means. There have emerged many techniques for protecting the security of multimedia information in recent years. Image digital watermark is a booming technique, which, in fact, is information hiding technique using in image processing field. The key to design a good watermark algorithm is invisibility and robustness.As an important branch of information hiding, we analyzed the current watermark developing status, basic characters and problems of digital watermark technology. The algorithm in frequency_domain is relatively robust, compared with the one in spatial_domain.This dissertation proposed two kinds of adaptive digital image watermarking algorithms in DCT and DWT by multi_embedding. The proposed algorithm is different from the methods in common digests. DC, mid_frequency and parts of low_frequency coefficients in DCT are selected as locations where to embeded a watermark. One watermark pixel is embedded many times (chip rate) into different DCT coefficients in same block. Not averaging watermark values extracted from one block, We use an intelligent method to get the last watermark. At the same time, watermark casting is performed by exploiting the masking characteristics of the Human Visual System.In DWT, we extended one watermark pixel into a block. It equals to embedding one pixel many times. Using the origial image texture,the watermark is performed. Then all coefficients in same sub_block in DWT are added. By judging the sign of the sum , we get the last mark.The invisibility and robustness were supported by our experiments.