An Effective Scheme for Copy-move Forgery Detection Using Polar Sine Transform

Nowadays, as rapid development of image processing technique, it is necessary to authenticate whether an image is an artificial image or not. Copy-move forgery is the frequently-used image manipulation means. In this paper, Polar Sine Transform (PST) and Locality Sensitive Hashing (LSH) are combined to detect copy-move forgery image. First, the detected image is divided into lots of overlapping blocks. And then, PST is applied to extract feature from each block. Subsequently, LSH is applied to classify these block features, and we search the similar features in same class as candidate block feature pairs. Finally, the post-processing operation using Euclidean distance is presented to filter out the weak block feature pairs. A series of experiments show that the proposed method has robustness to rotation and JPEG compression and is superior to the state-of-the-art methods.