针对现有的图像区域复制篡改检测算法其后处理的鲁棒性较差,且时间复杂度高的问题,提出了一种有效的检测与定位篡改区域算法.该算法通过对图像进行小波分解,将低频图像进行块分解,提取每块的Zernike矩特征,并将特征向量排序;然后,为每个特征向量搜索阈值符合的相似特征向量;最后,利用区域面积阈值去除错误的相似块对,并结合数学形态学定位篡改区域.实验结果表明,该算法不仅能有效地对抗常规的后处理操作,而且只在1/4图像上搜索块空间,提高了运算效率.本方法可用于面向图像内容的真实性鲁棒取证.%Most of the existing detection algorithms for testing the region duplication and forgery are not robust in the post region duplication image process, and have high time complexities. In this paper, an efficient and robust algorithm is presented for detecting and localizing this type of malicious forgery.After reducing the image dimension by Discrete Wavelet Transform(DWT), the Zernike moment is applied to the fixed sized overlapping blocks of a low-frequency image in the wavelet sub-band,and the eigenvectors are lexicographically sorted. Then, similar eigenvectors are matched by a certain threshold. Finally, the area threshold is proposed to remove the wrong similar blocks,and the mathematical morphology operations are performed to localize the forgery part. Experimental results show that proposed method is robust and not only can successfully detect this type of forgery for images subjected to post region duplication processing in various forms, but also can search block spaces in only the 1/4 image, which improves the calculation efficiency of algorithm. The method suits to the robust forensics for image contents.
展开▼
