Blind Forensics of Median Filtering in Digital Images

Abstract

Exposing the processing history of a digital image is an important problem for forensic analyzers and steganalyzers. As the median filter is a popular nonlinear denoising operator, the blind forensics of median filtering is particularly interesting. This paper proposes a novel approach for detecting median filtering in digital images, which can 1) accurately detect median filtering in arbitrary images, even reliably detect median filtering in low-resolution and JPEG compressed images; and 2) reliably detect tampering when part of a median-filtered image is inserted into a nonmedian-filtered image, or vice versa. The effectiveness of the proposed approach is exhaustively evaluated in five different image databases.