ABSTRACT
O.J. Lou, S.H. Li & Z.X. Liu Dalian University of Foreign Languages, Dalian, China
S.T. Tang Dalian University, Dalian, China
ABSTRACT: This paper presents a watermarking algorithm that is robust against geometric attacks by using a restoration based on Scale Invariant Feature Transform (SIFT) feature points. The algorithm extracts feature points set from the original image by using the SIFT. Before the watermark is detected, the algorithm can also extract new SIFT feature points set from the watermarked image. According to the Euclidean distance, it can identify matching points in the two feature point set, and then it uses the matched points to restore the images damaged by the geometric attacks. A watermark embedded in the low-frequency sub-band of contourlet, transforms by quantifying adaptively, and is extracted without the original image. Experimental results show that the algorithm is robust against conventional image processing attacks, geometric attacks, and combined attacks.
Keywords: Watermarking, feature point, image restoration, SIFT, contourlet, quantify
