Research On Image Watermarking Algorithm In DWT-Domain

With the development of computer and network, digital media can be spread and copied easily. At the same time, digital copyright protection has become a serious problem. Watermarking is considered as anew means to solve this problem. Information is embedded to host media to protect the integrity and the ownership.Compare with DCT that was applied in traditional image coding standard, wavelet transform has more advantages. Embedding watermark in wavelet domain has been researched in resent years.This thesis focuses on static image watermarking technology, providing a meaning watermark algorithm based on DWT. In order to enhance the security, the watermark data was scrammed with chaotic sequences before embedding. Because any trivial changes on initial value can lead to two different chaotic sequences, the watermark cannot be recovered without key. To improve the ability of resisting attack, the watermark data are embedded to both the low and high frequency coefficients of wavelet domain. Because the low frequency coefficients contains the most of the energy, Human Visual System and Genetic Algorithm are adopted to avoid the decreasing of the image quality, improving the invisible trait of the watermark. Only part of the original image information is involved in watermark detection, so it has practical meanings.Experimental results show that the proposed algorithm ensured the high quality of the image and robustness against some common image processing, such as cut, JPEG compress, noise attack, etc.