earticle

영어

A robust watermarking technique using wavelet transform/hybrid wavelet transform and SVD is proposed in the paper. Singular values of watermark are embedded in mid-frequency transform coefficients of host located in a single row. Selection of single row makes only 256 transform coefficients available where watermark singular values can be embedded. By using the compression property of Singular Value Decomposition, only few singular values are embedded in the host row. Transform coefficients of host are sorted in descending order and their index positions are used to embed singular values. Scaling factor is also adaptively selected based on highest transform coefficient in the selected host row and highest singular value of watermark. Selection of single row makes it difficult to predict the location where watermark is embedded thus making it more robust over selecting multiple rows for embedding. Adaptive scaling factor adds to the robustness by adjusting the strength of watermark according to the highest singular value of watermark and host row selected for embedding. This technique proves better in terms of robustness against various attacks than embedding singular values of watermark in multiple rows of host proposed in our previous work.