基于人类视觉模型和Contourlet变换的图像感知哈希算法  

Image perceptual Hashing algorithm based on human visual system and contourlet transform

作　　者：邢慧芬[1] 吴其林[1] 曹骞[1] 

机构地区：[1]巢湖学院信息工程学院,安徽合肥238000

出　　处：《阜阳师范学院学报（自然科学版）》2016年第4期62-66,共5页Journal of Fuyang Normal University(Natural Science)

基　　金：巢湖学院科研课题(XLY-201410)资助

摘　　要：图像感知哈希(Perceptual Hashing)技术在图像的认证、识别和检索中得到广泛应用。融合人眼视觉系统(HVS)、Contourlet变换及奇异值分解(SVD)提出了一种新颖的图像感知哈希算法。该算法首先对图像进行Contourlet变换,计算变换后系数的视觉掩蔽特征值(掩蔽矩阵);然后对掩蔽矩阵分块后作奇异值分解,取每块最大奇异值作为图像的特征值,经过量化编码、压缩,生成最终哈希。该算法使用MATLAB作为实验平台,实验结果证明算法对大部分的感知保持操作具有较好的鲁棒性,不同图像之间也有较好的唯一性,同时对哈希进行加密处理,使得算法具有良好的安全性。Perceptual Hashing is an emerging technology, which is widely used in image authentication, image identifica- tion and image retrieval. Integration of Human Visual System （HVS）, Contourlet Transform and Singular Value Decomposition （SVD）, a novel algorithm on Image Perceptual Hashing was proposed. Firstly, the method used Contourlet Transform to get the low-frequency coefficients of the image. Second,visual masking characteristic value was calculated from the coefficients of the image. Third, masking matrix was partitioned and each block of masking matrix used Singular Value Decomposition to get larg- est singular value of each block as a feature value of the image. Finally, the final hash was generated by the quantized coding and compression. The MATLAB was used as an experimental platform, and the experiment results showed that the algorithm had better robustness for most of perception holding operations, strong uniqueness of the differences between the images. Be- sides, the hash was encrypted so that the algorithm had good security.

关 键 词：感知哈希 人类视觉系统 CONTOURLET变换 奇异值分解 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]