Automatic Forgery Localization via Artifacts Analysis of JPEG and Resampling

Abstract

With the availability of highly sophisticated editing tools, the authenticity of digital images has now become questionable. The level of image tampering is getting higher and higher, and the tampering procedures become more and more complicated. To recognize the tampering area of the original image, the tampered image is usually executed a series of post-processing. This behavior has greatly increased the difficulty of forgery detection. In this paper, a blind JPEG image forgery detection and localization technique based on JPEG and resampling artifacts analysis is proposed. The process of tampering is to first tamper with JPEG images by bitmaps. Then original JPEG image and tampered area are manipulated by a series of operations, that is, the image is enlarged and then saved as JPEG. A novel tampering localization method is presented based on resampling and JPEG blockness artifacts. Theoretical analysis and experimental results show that the proposed method can effectively identify and locate the tampered region of a spliced image with a JPEG-resampling-JPEG operation chain.