基于DWT-SVD的可伸缩图像渐进性水印算法  

Progressive Watermarking Algorithm of Scalable Image Based on DWT-SVD

作　　者：陈青[1] 熊蒙 周广州[1] 

机构地区：[1]上海理工大学,上海200093

出　　处：《包装工程》2017年第3期129-133,共5页Packaging Engineering

基　　金：上海自然科学基金(12ZR1420800);国家863计划(2012AA050206)

摘　　要：目的提出基于DWT-SVD的可伸缩图像渐进性水印算法,可用于可伸缩图像的版权保护,对可伸缩图像的安全技术具有重要的意义。方法对载体图像进行三级小波分解,确保能量的集中,减小因攻击对水印信息的影响;对各个小波分解的子带都进行H×H不重复分块,对每个子块进行奇异值分解,并计算第1个奇异值,以便选取合适的嵌入强度,经综合考虑,该算法中嵌入强度选为5;将灰度水印图像进行8个位平面分解,根据重要性降低的次序依次嵌入到载体图像三级小波分解和量化后的重要性排序降低的分辨率子带中。结果压缩比从90~20,随着压缩比的减小,提取灰度水印图像的位平面增加,图像质量不断改善,NC值逐渐增大。当压缩比在70以下时,对最低频子带系数无影响;当压缩比增大至80以上时,对最低频子带信息产生一定的影响,但水印图像依然可以提取。不同攻击和常规的信号处理攻击后各级分辨率均可提取到水印信息,水印质量随着分辨率增加逐渐变好。结论渐进性提取和鲁棒性仿真实验表明该算法能很好地抵抗压缩攻击和剪切攻击,并对常规的信号处理攻击具有很好的鲁棒性。The work aims to put forward the progressive watermarking algorithm of scalable image based on DWT-SVD, which can be used to protect the copyright of the scalable image and has great signification for security technology of the scalable image. A three-level wavelet of carrier images was decomposed to ensure the energy concentration, thus reducing the impact of attacks on watermark information. The H×H non-repetitive blocking of sub-bands decomposed by each wavelet was conducted. The singular value of each sub block was decomposed. Then the first singular value was calculated to help select the proper embedding strength. Through comprehensive consideration, the embedding strength of the algorithm was 5. Finally, according to the descending order of importance, the eight bit-plane decomposition of the gray scale watermark image was done to embed it into the three-level wavelet decomposition of carrier image and into the resolution ratio sub-band after quantization. Compression ratio decreased from 90 to 20. With the decrease of compression ratio, the bit plane extracted from gray watermark image was increased, thus image quality was improved and NC value was increased gradually. When the compression ratio was below 70, it would have no influence on the lowest frequency sub-band coefficient. When the compression ratio was increased to above 80, it would have certain effect on the lowest frequency sub-band information, but the watermark image could still be extracted. After different attacks and regular signal processing attacks, watermark information could be extracted from resolution ratios at different levels, while the watermark quality would become better with the resolution ratio increased. The progressive extraction and robustness simulation experiment indicate that this arithmetic can well resist the compression attack and shear attack and it also has better robustness of regular signal processing.

关 键 词：JPEG2000 小波 渐进提取 信号处理攻击 

分 类 号：TS801.3[轻工技术与工程] TP391[自动化与计算机技术—计算机应用技术]