基于小波系数相关性的图像篡改盲检测及定位  

Blind image tampered detection and localization based on correlation of wavelet coefficient

作　　者：张娜[1] 王阿川[1] 

机构地区：[1]东北林业大学

出　　处：《计算机工程与应用》2013年第11期199-203,共5页Computer Engineering and Applications

摘　　要：分析了小波变换模极大值与信号突变点之间存在的联系,以及图像信息相关性的强弱,实现对图像的篡改部分进行检测。由于对图像进行篡改操作之后,通常会在两幅图像的结合处出现突变点(或者奇异点),形成强边缘。因此,利用小波变换的多分辨率分析,求得模极大值进行多项式曲线拟合的结果,根据拟合误差的大小对图像的强边缘进行检测定位,再通过图像分块之间相关性的强弱,对篡改区域进行更准确的篡改检测和定位分析。实验结果证明,该方法能够检测到篡改信息,并且可对图像中的篡改部分进行准确定位。Based on analyzing the contact between the modulus maximum of wavelet transform and the mutations points of the signal, and the strength of the correlation in the image information, the tampered areas of the images are detected. Due to operat- ing the image on tampering （splicing）, mutation points （or singular points） usually appear in the department of two images, and thus the strong edges are formed. Therefore in this paper, by using the multiresolution analysis of wavelet transform, it obtains the modulus maxima of wavelet coefficients in each layer, and fits the curve, thereby according to the error＇ s size of the fitting, the strong edges of the images are detected and located, through computing the size of the correlation coefficient between the image blocks again, so as to detect, locate and analyze the tampered areas in the tampered images further. Experimental results show that, this method can detect the strong edges, and the tampered areas in the tampered images are detected and located accurately.

关 键 词：图像篡改检测 小波变换 多分辨率分析 突变点检测 相关系数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]