Key point descriptor based copy and move image forgery detection system

Abstract

In recent years Image forgery plays important role in many areas like Internet, Forensic department, Passports, License, and Educational certificates etc. Several types of Image forgeries are available such as copy-move, copy-paste, cut-copy. In our work we focused on copy-move image forgery. To identify the forgery region in an image we proposed the adaptive over segmentation and key point matching algorithm. Lots of researchers already proposed algorithm for copy move forgery scheme. But still now the computational complexity is high. It can be reduced in our work by dividing image into non overlapping blocks of image region. The blocks of SIFT features are matched with neighboring block features and locate the forgery region. The experimental shows that we achieved high recall rate under different transform (DCT, SVD, FMT).