基于HVS感知模型与数据压缩的可逆图像水印算法  

Reversible Image Watermarking Algorithm Based on HVS and Data Compression

作　　者：李书欣[1] 王泉[1] 吴延赞[1] 刘梦楠[1] 

机构地区：[1]西安电子科技大学计算机外部设备研究所,陕西西安710071

出　　处：《电视技术》2010年第6期26-27,37,共3页Video Engineering

基　　金：国家大学生创新性实验计划项目(091070114)

摘　　要：提出了一种基于人类视觉系统(HVS)感知模型和数据压缩技术可逆图像水印算法。算法首先将原始图像进行分块,每块进行DCT变换,利用JND模型对变换域系数进行处理,得到每个像素最大可提取的有效位并提取。将提取出的有效位串进行压缩,得到压缩位串,将水印信息与压缩位串连接在一起嵌入原始图像,同时记录边信息,包括每个像素点的最大可提取有效位数目以及嵌入位置,提取时可据此边信息完全恢复原始图像,并得到水印信息。容量的大小取决于所使用的压缩算法,使用游程编码进行压缩。同时,运用人类视觉系统的感知模型,保证了嵌入水印后的图像的视觉质量。In this paper, a reversible image watermarking algorithm based on Human Visual System （HVS） and data compression is presented. The original image is firstly divided into non-overiapping blocks, to which the DCT is applied. After the modulation of the DCT coefficients by the JND model, the suitable least significant bits are generated. The compressed bits are concatenated with the watermarks and embedded into the image. Meanwhile, the side information including the least significant bits of the pixels and the embedding locations is recorded, which is necessary in the extraction process. The capacity of the algorithm is depended on the compression technique, so run-length coding is used in this paper. By utilizing the HVS, the visual quality of the watermarked image is guaranteed.

关 键 词：DCT 人类视觉系统 JND模型 游程编码 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]