Adaptive matching for copy-move Forgery detection

The objective of copy-move forgery detection methods are to find copied regions within the same image. There are two main approaches to detect copy-move forgery: keypoint-based and block-based methods. Although the former is superior in terms of computational complexity, these methods neglect the smooth regions since they confine their search to salient points. On the other hand, while block-based methods consider smooth areas, they introduce a huge number of false matches. In this paper, it is proposed to employ an adaptive threshold in the matching phase in order to overcome this problem. The experimental results demonstrate that the proposed method can greatly reduce the number of false matches which results in improving both performance and computational cost.