一种新的视觉Hash算法  

A novel perceptual image Hash algorithm

作　　者：赵玉鑫[1] 刘光杰[1] 戴跃伟[1] 王执铨[1] 

机构地区：[1]南京理工大学自动化学院,江苏南京210094

出　　处：《光学精密工程》2008年第3期551-557,共7页Optics and Precision Engineering

基　　金：国家自然科学基金资助项目(No.60374066,No.60574082);江苏省自然科学基金资助项目(No.BK2004132)

摘　　要：提出了一种新的视觉Hash方案。利用密钥提取图像鲁棒特征,并对特征进行Hash运算。通过在密钥的生成过程中引入图像独特的光学特性,增强Hash的安全性。实验结果表明,本方案对JPEG压缩、滤波、噪声等处理具有良好的鲁棒性,Hash误比特率低于0.01,而对图像的恶意篡改有较强的敏感性,误比特率在0.5左右。本方案可用于面向对象的真实性鲁棒认证。In order to achieve content authentication,a novel perceptual image Hash scheme is proposed.The robust feature of an image is extracted by secret key,and its image Hash is computed by the feature generation.To enhance the security of Hash,the image optical feature is introduced in the creation of secret key.The experimental results show that the proposed scheme is robust for JPEG compression,filtering,noise and some else managements,with the Hash bit error rate of less than 0.01; but it is fragile against t...

关 键 词：视觉Hash算法 光学密钥 透镜畸变 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]