| Summary: | 碩士 === 國立中興大學 === 通訊工程研究所 === 98 === This paper proposes a informed watermarking based on trellis structure, and divided into informed coding and informed embedding . We present two fast informed embedding algorithms for watermarking system. These algorithms are capable of embedding 1585 bits of message in an gray-level image of size 512X512 pixels. These algorithms employ the codewords of linear block code to label the arcs in the convolutional trellis. With this characteristic, we can easily adjust the fidelity and robustness of watermarked images by some controllable parameters. We report the robustness and complexity performance of these algorithms under the fixed fidelity in comparison with Miller''s. And we show the robustness results in Gaussian noise, lossy compression, low-pass filer and valumetric scaling. Moreover, the proposed algorithms not only has less complexity but also have better BER performance compared with Miller''s.
|
|---|
