A blind watermarking algorithm resisting to geometric transforms based on SVD

A new watermarking algorithm resisting to geometric transformation based on singular value decomposition (SVD) in logarithm polar coordinate is proposed. The log-polar mapping (LPM) is used to resist rotation and scaling attacks, and the odd-even quantization algorithm is used to embed watermark so it can be extracted without the original host image. The experiments show that the proposed algorithm not only resists various geometric attacks but also is robust enough to the common signal processing.