A fast DCT based method for copy move forgery detection

Copy move forgery detection is emerging as one of the hot research topic among researchers in the area of image forensics. Many techniques have been suggested to detect such type of tampering with the original image, but the problem is far from being solved. Some issues still remained either unsolved or there is a lot of scope for performance improvement. Block matching algorithm or block tiling algorithm is the most commonly used method to detect the duplication in the image. One of the major challenges is the time complexity of such algorithms. As the image size increases the number of overlapping blocks increases rapidly and feature collection and matching takes relatively long time. In the proposed method this issue has been addressed without compromising the quality of the method. Discrete Cosine Transform (DCT) is used to represent the features of overlapping blocks. Also effort has been made to automate the threshold for separating the stray matches from the authentic matches.