特征增强的平滑区域篡改检测算法

Detection of Smoothing Region Tampering Based on Feature Enhancement

针对移除图像平滑区域目标的篡改行为,提出一种基于轮廓特征增强的检测算法.首先对待检测图像进行滑窗式分块,逐块进行二值化轮廓特征提取,然后对提取的轮廓序列进行字典排序,最后通过对轮廓序列的匹配实现篡改区域的正确标注.实验表明该算法能有效解决平滑区域篡改痕迹检测的问题,且篡改图像经较低品质JPEG压缩后仍有较强的匹配能力.

An algorithm for forgery detection based on outline feature enhancement is proposed to remove tampered areas in an image. The image to be detected is first divided into blocks of sliding-windows. Binary outline features are extracted from every block. The extracted outline sequences are ranked in an alphabetic order. By matching the outline sequences, tampered regions are identified. Experiment results show that the proposed algorithm can effectively detect tampering traces in smooth image regions, even from images after JPEG compression with low quality factors.