基于奇异值分解的遥感图像鲁棒水印算法探讨  

A robust watermarking algorithm based on DCT-SVD for remote sensing images

作　　者：侯翔[1] 杨辉[1] 

机构地区：[1]信息工程大学地理空间信息学院,郑州450052

出　　处：《测绘科学》2015年第7期123-126,158,共5页Science of Surveying and Mapping

摘　　要：针对遥感图像的安全保护问题,文章提出了一种基于DCT-SVD和最近区间量化的鲁棒水印算法,将遥感图像进行分块,选择视觉掩蔽效应好的纹理子块作为嵌入区域,根据混合变换后的系数在各自所属区间上的位置,将系数进行最近区间量化来嵌入水印信息;利用密钥即可实现水印的提取,不需要原始遥感图像参与。实验结果表明,该算法对加噪、滤波、旋转、缩放、随机剪切、JPEG压缩等攻击均具有较强的鲁棒性,可有效保护遥感图像的安全。Aiming at the problem in the security protection of remote sensing images, this paper proposed a robust watermarking algorithm based on DCT-SVD and nearest interval quantizing. Remote sensing image was divided into blocks, and texture blocks which have better visual masking effect were selected as the embedding zones, based on the location of interval each conversion coefficient belongs to, the coefficients were performed nearest interval quantizing in order to embed watermarking information. Moreover, the watermarking extraction was achieved by secret key without resorting the original images. Experiment results showed that high robustness could be obtained under the attacks of adding noise, filtering, rotating, scaling, random cropping, and JPEG compression, etc., which proved its validity in the security protection of remote sensing images.

关 键 词：遥感图像 离散余弦变换 奇异值分解 纹理子块 最近区间量化 

分 类 号：P237.9[天文地球—摄影测量与遥感] TP75[天文地球—测绘科学与技术]