Study On Watermark Algorithms Base On Wavelet

In this paper, we mainly describe the digital watermarking technologies base on wavelet domain. We firstly introduced the basic principles, the basic capability and evaluation of performance of the digital watermarking technology. Then we adopted some new technologies such as wavelet transform, random number of flag sequence, SVD (Singular Value Decomposition), image-adaptive, video frame, into the digital watermarking and proposed several different kinds of watermarking schemes: watermarking in binary image; watermarking of adopting flag sequence in the color image; visible watermarking of image-adaptive; watermarking base on wavelet transform and SVD domain in video. The proposed schemes in this paper could cover the primary aspects of the image digital watermarking, cover images are likely to color, gray, static or video; watermarkings are likely to binary, gray, visible or invisible.Cove image and watermark image are binary images. Cove image is divided into 8~*8 block, computing the rate of the black dot and white dot in every block. Watermark is embedded or distilled in binary image according to adjusting the rate.Cove image is color and watermark is binary. Cove image is transformed into YCrCb space, then it is transformed into wavelet and divided into blocks. Computing the mean gray of every block, watermark is embedded or distilled into gray image according to random number of flag sequence.Cove image and watermark image are all gray. Cove image and watermark image are all transformed into wavelet, then low frequency of watermark is embedded into low frequency of cove image according to gray mask adaptively and high frequency of watermark is embedded into high frequency of cove image according to high frequency characteristic adaptively.Cove image is gray videos and watermark is gray image. Videos are transformed into one dimension wavelet according to time. Then watermark is embedded or distilled the coefficient of low frequency which are transformed into SVD(Singular Value Decomposition).