Abstract
For the problem of copyright protection and ownership identification, the digital watermarking technique was brought up. The watermarking strategies can partially solve this problem. In this thesis, we apply a wavelet-based watermarking algorithm. We adopt the 5/3 and 9/7 wavelet transform to the original image and a binary pattern as our watermark. The binary pattern is a contour which shows the boundary of objects. The watermark is embedded into the coefficients of the high-low band and the low-high band at level 2 in a 2-scale wavelet transform of the original image. A watermark acquisition, embedding, extraction, and detection after a variety of the attacks by image processing operations and lossy data compression such as scaling, smoothing, cropping, noise adding, JPEG compression, and JPEG2000 compression are demonstrated. The PSNR values and the ratio of watermark detection of the attacks in various experiments show that our proposed approach is promising.