Detection of Double-Compression in JPEG2000 Images for Application in Image Forensics

Abstract

In the age of digital multimedia, the emergence of high resolution digital cameras and sophisticated photo-editing software packages makes it relatively easy to create digital image forgeries. It is increasingly important to verify the authenticity and integrity of digital images. In this paper, a new approach to detect tampered images based on double JPEG 2000 compression is proposed. The paper first analyzes the statistics of double JPEG 2000 compression, the artifacts in the Fourier transforms of DWT (Discrete Wavelet Transform) coefficient histogram, i.e., the stronger singularity in the mid and high frequencies compared with single JPEG 2000 compression; then extracts the singularity using the characteristics of DWT and detects double JPEG 2000 compression by thresholding the singularity. Finally, to images identified as the double JPEG 2000 compressed versions, the percent of non-zero DWT coefficients in high frequencies is examined to catch the difference between authentic image part and tampered image part. The experiments demonstrate that the proposed approach can achieve an effective detection for double JPEG 2000 compression as well as accurate location for the tampered areas.