基于自组织竞争神经网络和小波变换的密写  

Steganography Based on Self-organizing Competitive NNs and Wavelet

作　　者：黄若宏[1] 刘怀兰[2] 戴军[3] 

机构地区：[1]华中科技大学计算机科学与技术学院,湖北武汉430074 [2]华中科技大学机械科学与工程学院,湖北武汉430074 [3]吉首大学现代教育技术中心,湖南吉首416000

出　　处：《小型微型计算机系统》2009年第8期1685-1688,共4页Journal of Chinese Computer Systems

基　　金：国家科技型中小企业技术创新基金项目(04C26224210717)资助

摘　　要：依照人眼视觉特性对边缘和较暗区域不敏感的特点,提出基于自组织竞争神经网络和小波变换的图像密写方法.该方法将载体图像分成若干固定大小的子块,对各子块进行一级小波分解后计算小波对比度和大幅值小波系数的数量,利用自组织竞争神经网络将各子块分为三类,并采用模算子技术嵌入秘密信息.实验结果表明,该方法在较好地保持原载密图像视觉效果的同时可以嵌入更多的秘密信息.In accordance with the property of Human Visual System （HVS） that is insensitive to edged and dark areas, a novel steganographic solution based on self-organizing competitive NNs and wavelet is proposed. An image is divided into many small blocks. Transform each block into one-level wavelet in order to obtain the wavelet contrast and the count of nonzero wavelet coefficients. Classify all blocks by self-organizing NNs and embed secret information with steganography based on modulus. The experiments result in that the proposed method can hide much more information while the vision of original stego-image is good remained.

关 键 词：密写 小波变换 小波对比度 自组织竞争神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]