多特征视觉Hash方案

Perceptual image hash scheme based on multiple features

提出一种基于认证安全性的视觉Hash设计方案。通过对视觉Hash认证中鲁棒性、认证集合相互关系的分析,指出了采用多特征构造视觉Hash的必要性。多特征方法可以平衡视觉Hash认证中的虚警、漏警问题。设计了一种基于小波分解的视觉Hash算法。对多次分解的小波低频系数进行量化提取多重特征,利用精确Hash算法组合生成认证Hash。实验结果表明本方案对JPEG压缩、滤波、噪声等处理有良好的鲁棒性,且具有较好的认证安全性,可以用于图像的真实性认证。

A perceptual image hashing scheme based on authentication security is proposed. The security analysis proves that multiple features are necessary in perceptual image Hash scheme. The false negative probability and false positive probability can be balanced in authentication by using multiple features. A novel perceptual image hash algorithm is designed based on wavelet transform. Multiple features are extracted and quantized after multiple wavelet segmentations. Then the precise Hash algorithm is used to generate authentication Hash. Experimental results show that the proposed scheme is robust a gainst JPEG compression,filtering,noise and some else managements, It can satisfy the multimedia content authentication.