Singular value decomposition and space/spatial-frequency analysis applied to digital watermarking
Nikola Žarić, Irena Orović, Srdjan S. Stankovic · European Signal Processing Conference · 2012
A watermarking procedure in the space/spatial-frequency domain, based on the singular value decomposition is proposed. Singular values are used for selection of pixels suitable for watermarking. Generally, regions for watermarking are characterized as: busy and moderate busy. For pixels from busy regions the entire middle frequency content in the space/spatial frequency domain is used for watermarking. In the case of moderate busy pixels the left and right singular vectors are used to shape watermark embedding region. Watermark modeling, embedding, and detection are performed in the space/spatial-frequency domain. The efficiency of the proposed procedure is tested under various attacks.