Statistical Detector of Resampled TIFF Images
Abstract
Resampling is a common manipulation in creating tampered digital images. The goal of this paper is to develop an efficient detector to distinguish between a resampled TIFF image from an original TIFF image. To this end, we first propose a statistical model for resampled TIFF images by analyzing the complete processing process from a RAW image to a resampled TIFF image. Next, we formulate the detection problem as a likelihood ratio test between the models of original and resampled TIFF images. The test power is analytically evaluated in the context that all model parameters of original TIFF images are unknown. Numerous numerical experiments justify the performances of the detector.