失真补偿量化索引调制水印的性能分析  

Performance Analysis of Distortion-Compensated QIM Watermarking

作　　者：曾高荣[1,2] 裘正定[1] 章春娥[1] 

机构地区：[1]北京交通大学信息所,北京100044 [2]上饶师范学院大计部,上饶334001

出　　处：《电子与信息学报》2010年第1期86-91,共6页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(60773015);博士学科点专项基金(20060004007);北京交通大学校基金(2007RC029)资助课题

摘　　要：该文分析了基于量化索引调制的失真补偿水印方案中量化步长的伸缩因子与失真度、鲁棒性和检测误码率之间的关系。在高分辨率量化的假设前提下,推导出失真补偿量化索引调制水印算法在加性高斯白噪声信道下的误码率计算公式,并提出了一个失真-鲁棒性函数来度量算法的抗噪鲁棒性和失真度与鲁棒性之间的代价关系。通过计算机仿真,比较了不同失真补偿量化索引调制水印方案下算法的失真-鲁棒性能和误码率性能;并以一段音频数据三级小波细节系数为载体嵌入水印信息,统计检测时的位错误率。实验结果验证了在低量化步长范围内,理论预测误码率与实验检测误码率较为吻合,失真-鲁棒性函数能够很好地度量高分辨率量化前提下的算法抗噪鲁棒性和失真度与鲁棒性之间的代价关系。Concerning the distortion, robustness and decoding error probability, this paper analyses the performance of Distortion-Compensated Quantization Index Modulation （DC-QIM） watermarking scheme with a scaling factor of step. On the assumption that quantization is high resolution, a formula of decoding error probability for DC-QIM watermarking scheme is derived under the Additive White Gaussian Noise （AWGN） channel, and a distortion-robustness function is defined as a criterion characterizing the tradeoff between distortion and the robustness. In the experiment, several audio sections are selected as the hosts and their third lever wavelet detail coefficients are quantified according to watermark bit series. In term of the distortion-robustness and decoding error probability, which of different DC-QIM schemes are compared, simulation result and experimental value are in accord with predictive one when step is on the small side.

关 键 词：数字水印 失真补偿量化索引调制 失真-鲁棒性 误码率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]