关键词: 篡改检测, 傅里叶－梅林变换, 数字图像取证, 区域复制篡改

Abstract: Tamper detection has become an important component of digital image forensics and multimedia security.In most of the time，digital image tamper is not perceptible by human，however，some traces of digital tamper may be left in the image during the manipulation.This paper proposes a novel scheme to detect and identify the location of image copy-move tamper in which an object region in the image is copied and pasted somewhere else in the same image to confound right and wrong.The Fourier-Mellin Transform（FMT） serve to produce a translation，rotation and scale invariant of local image patches，suspected to be a copy，and cosine correlation coefficient is considered for similarity metric.Experimental results also demonstrate the efficacy of the proposed approach to tampered images by geometric transformations（i.e.translation，rotation and scale） undergone some attacks like blur filter，random noise contamination，lossy JPEG compression，etc.

Key words: tamper detection, Fourier-Mellin Transform（FMT）, digital image forensics, copy-move