Adaptive segmentation-based feature extraction and S-STDM watermarking method for color image

Abstract

A local color image watermarking scheme using adaptive segmentation-based feature extraction (ASFE) and singular value decomposition-based spread transform dither modulation (S-STDM) is proposed in this paper. The proposed ASFE can adaptively extract feature regions from blocks segmented by the simple linear iterative clustering. In addition, the stationary wavelet transform is employed to decompose each of the extracted feature regions because of its shift invariance. Consequently, the novel S-STDM watermarking method is proposed, where we employ singular value decomposition to calculate the approximation coefficients and then select the generated diagonal elements from the decomposed diagonal matrix for watermark message embedding. The experimental results show that the proposed scheme is superior to the existing schemes under various attacks.