Detection of copy-move forgery image using Gabor descriptor

Abstract

In this paper a new forgery detection system which uses a novel feature extraction and located algorithm is proposed. The Gabor feature of image is used to detect the forged image caused by copy and move operations. The Gabor filter with different scaling factors, rotation angles and frequencies are considered to generate the Gabor feature representation of an image. For comparing two images, their Gabor features are applied to find if there is any similarity between them. To reduce the processing time and detect the small copy and move area, an image is divided into small blocks. In each block, it extracts a new descriptor and key points from the Gabor feature of the block image. Furthermore, the proposed method can also locate the duplicated regions precisely and detects the rotation angles and scale factors of the forged image. Experimental result shows that the proposed method can achieve high correct rate, estimated rotation angle and scaling factor. The proposed method has robustness and reliability.