An Efficient Ward-Based Copy-Move Forgery Detection Method for Digital Image Forensic

In recent years, with the improvement of digital manipulation tools, image forgeries have become a genuine social issue. Copy-move forgery is an important type of digital manipulation which is done by duplicating some parts of the image with the purpose of hiding specific information. Verifying the authenticity of the digital image can become vital for different forensic organizations. In this paper, an efficient copy-move detection algorithm with three level of Ward linkage based clustering is proposed. The proposed algorithm extracts the local image features with the help of Scale-invariant feature transform algorithm (SIFT). In the proposed method, a Ward-based clustering scheme followed by three levels of false-descriptor elimination is applied. The proposed elimination system utilizes Euclidean distance to improve the accuracy of the region detection scheme. The efficiency of the proposed algorithm is examined against several different post-processing attacks such as multilevel forgeries, noises, and JPEG compression. Experimental results illustrate that the proposed algorithm improves the detection accuracy by eliminating false descriptors.