中文摘要：

提出了一种基于小波变换和奇异值分解的盲检测算法来识别图像的复制粘贴伪造。该算法用小波变换降低计算量,用奇异值表示图像特征。图像经过小波变换,提取出低频分量和高频分量。因低频部分保留图像的纹理信息,高频部分保留图像的轮廓信息,该算法分别从低频部分和高频部分提取图像奇异值,并把提取出的奇异值进行加权处理,以加权值作为图像块的特征。图像块之间做两两比较,根据图像块的特征相似度,判断是否存在图像复制粘贴伪造区域。在丰富层次和清晰细节轮廓的图像中,该算法能达到比较理想的检测效果,准确率较高。

英文摘要：

An efficient and passive-blind approach based on DWT and SVD is presented to detect and identify the location of copy-region forgery in a digital image.Firstly,the algorithm applies DWT to the image,and the low-frequency component and high-frequency component are extracted,divides the image into many small blocks with appropriate size.The singular values extracted from low-frequency component and high-frequency component by SVD are weighted as the features of block.By comparing the features block by block,the copy forgery regions are localized.Experiments demonstrate that the copy forgery regions are detected accurately in the image with rich details and clear outline.