A set of coefficient selection rules is proposed for efficiently determining the effective DOT watermarking coefficients of an image for JPEG attack. These rules are simple in computation but they are derived from the theoretically optimized data set with the aid of the parametric classifiers. They improve the watermark robustness (correctly decoding) and, in the mean time, decrease the error detection probability (correct detection). The frequency versus watermark strength space is used in constructing the selection rules. Simulation results show that the computational complexity is significantly reduced compared to our previous theory-based optimization work, but still the selected coefficients can achieve nearly the same performance as the original scheme.
|頁（從 - 到）||763-770|
|期刊||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|出版狀態||已出版 - 2004|