Image Forgery Detection and Localization using DCT-based Forensic Analysis Approach

Abstract

Digital image tempering is widespread because of software and devices that manipulate image information without any traces are easily available for high-performance image editing. Now everything is online for a day and digital images are presented as evidence of any event, documentation where forgery hides its traces. Existing techniques for forgery detection are based on the higher complexity of computational costs. The technique proposed is robust even with pre- and post-processing operations for automatic detection and localization of specific artifacts. We proposed a DCT supported technique to obtain features from each block of images that reduce the block dimension and assist in lexicographic sorting. Tampered blocks of images are compared with predefined threshold values based on robust parameters to detect similar blocks in reduced time. Experimental results are robust to multiple forgeries with low computational complexity and retained significance of image information. On several images that are affected by different forgery types, we show our method robustly.