Abstract
The problem of copyright protection and ownership identification is attention-getting because the Internet applications and consumer electronics products have been rapidly developed. The watermarking system can partially solve this problem. In this thesis, experiments of a watermarking system using 4 different discrete wavelet transforms were implemented. An authorized watermark, sampled from a standard normal distribution, is embedded into the wavelet coefficients of high-low and low-high subbands at level 3 of a 3-scale discrete wavelet transform. Four popular wavelet bases including Haar, Daubechies’ four, 5/3 and 9/7 are chosen for experiments. Our experiments, including scaling, median filtering, cropping, smoothing, noise adding, JPEG compression and multiple watermarks, demonstrate that our watermarking scheme satisfies fidelity, robustness, and security at a certain level.