A Scaling Robust Copy-Paste Tampering Detection for Digital Image Forensics

Abstract

It is crucial in image forensics to prove the authenticity of the digital images. Due to the availability of the using sophisticated image editing software programs, anyone can manipulate the images easily. There are various types of digital image manipulation or tampering possible; like image compositing, splicing, copy-paste, etc. In this paper, we propose a passive scaling robust algorithm for the detection of Copy-Paste tampering. Sometimes the copied region of an image is scaled before pasting to some other location in the image. In such cases, the normal Copy-Paste detection algorithm fails to detect the forgeries. We have implemented and used an improved customized Normalized Cross Correlation for detecting highly correlated areas from the image and the image blocks, thereby detecting the tampered regions from an image. The experimental results demonstrate that the proposed approach can be effectively used to detect copy-paste forgeries accurately and is scaling robust.