Comparison of Watermarking Schemes Using Linear and Nonlinear Feature Matching

Recently, the feature point matching based water-marking techniques have been paying attention for resisting the geometric attacks. We present a performance analysis of robust watermarking using linear and nonlinear feature. In particular, we consider the geometric attacks and the signal processing attacks for image watermarking. In order to analyze the efficiency of linear and nonlinear feature, we employ the linear and the nonlinear feature matching technique in the image watermarking. The extracted feature points can survive against several attacks, therefore, those can be used as reference points for restoration before the extraction of the watermark information. For blindness and robustness, we embed the watermark into the low-band of the discrete cosine transform (DCT) domain. Experimental results show that our performance analysis of watermarking methods using linear and nonlinear feature matching against the geometric attacks and the signal processing attacks. These include the JPEG compression, the filtering attacks, and so on.