Digital Image Forensics Based on Local Energy Variance Properties

QIAO Tong,QIAN Zhen-Xing,ZHANG Xin-Peng,WANG Wen-Wen

DOI: https://doi.org/10.3969/j.issn.1003-6059.2012.02.014

2012-01-01

Pattern Recognition and Artificial Intelligence

Abstract:How to identify photographic images and computer graphics efficiently is more and more important in image forensics.An authentication algorithm for digital images is proposed using the correlations between photographic images and computer graphics.The calculated local energy dithering in the frequency domain is used to distinguish photographic images and computer-generated images.Moreover,compared with Gallagher's method,the proposed method is still available when both kinds of images above are attacked by CFA interpolation or zooming operation.The experimental results show that the proposed algorithm identifies better with the capability of anti-forensics resistance.