Detection of Image Forgery for Forensic Analytics

Abstract

Due to the technical revolution in digital image processing, different advanced image manipulation software has been used in recent years to produce new unrealistic images without leaving evidence of what is happening in the world, so it would be difficult to detect tampering visually. Digital image forgeries have many techniques, but it is still very difficult to identify copy-move forgery. Therefore, we use a robust algorithm in this paper to detect copy-move forgery based on the descriptor speed-up robust feature (SURF) as a key-point detection, high-pass filtering as a matching feature, nearest neighbor used as a clustering algorithm to divide the entire image. By swapping the matched feature points with the corresponding super pixel blocks, the doubtful regions are identified, and then, the corresponding blocks are combined on the basis of similar local color features (LCF). Finally, to obtain the suspected forged areas, morphological close operation was applied. The results of the study indicate that the proposed method achieves considerable output based on key-point detection compared to other forgery detection methods used in the current method in order to address the research challenges.