Search

This paper proposed a detection method of image copy-move forgery based on the interest points detection and feature matching. First, it applied Harris operator to detect corners as interest points, then extracted five spacial average features in the neighborhood of which as the center to constitute feature vector, and finally recorded the occurrence frequency of position transfer vector to obtain matching interest points with thresholding processing, thus marked copy-move regions. Experimental results show that the algorithm can not only detect multi-region copy-move manipulation effectively, but also be able to effectively resist various tampering post-processing operations, including additive white Gaussian noise, JPEG compression, contrast, brightness and exposure adjustment and the hybrid operation of JPEG compression and noise addition. [ABSTRACT FROM AUTHOR]