基于灰度共生矩的图像区域复制篡改检测

Detection of image region-duplication forgery based on gray level co-occurrence matrix

针对图像区域复制—粘贴篡改,提出了一种基于灰度共生矩阵的检测算法。首先将待检测图像分成大小相同的多个重叠块,用灰度共生矩阵的统计量表示每块图像的纹理特征,得到图像的特征矢量。然后将特征矢量进行字典排序,并结合图像块的位移矢量,检测且定位出篡改区域。实验结果表明,该算法在抗旋转处理和效率方面均优于经典的基于主成分分析法(PCA)的检测算法。

With regard to the copy-move forgery of image region,this paper proposes a detection algorithm based on gray level co-occurrence matrix.Firstly,we divided the detected image into multiple overlapping blocks with same sizes,represented the textural features of each block with the statistics of its gray level co-occurrence matrix,and got the feature vector of the image.Secondly,we sorted the feature vector by dictionary,and located the tampered region by utilizing the displacement vectors of image blocks.Lastly,experimental results show that our algorithm performs better than the classical detection algorithm based on Principal Component Analysis（PCA） in terms of the processing against rotate operation and of efficiency.