Detection and Localization of Copy-Move Tampering Using Features of Locality Preserving Projection

Abstract

The power of communication corresponding to visual content makes digital images a powerful and effective tool for developing idea and spreading messages. The digital image tampering is done with an intention to mislead the watchers feedback and create the false notion about the original image. One of the commonly used digital image tampering approach is copy-move tampering, in which desired part of image is replicated and undesired part is concealed in the same image. To make the tampering convincing various post-processing operations are performed on copy-move tampered images. In this paper the Locality Preserving Projection (LPP) based copy-move tampering detection scheme is proposed for the detection of copy-move tampering in the presence of such adversely operations. In our approach we are taking advantage of similarity preserving property of LPP. This approach effectively detects and localizes tampering in images with the presence of JPEG compression, additive Gaussian noise, colour reduction and brightness change. We can localize the tampered region even if edges of the pasted region is blur.