An actual blind detection watermark approach based on CT and ICA

This paper proposes a actual blind detection watermark scheme based on Contourlet Transform (CT) and Independent Component Analysis (ICA). To decrease Contourlet coefficient values of the watermark images and ensure safe of watermark, watermark images are encrypted. The encrypted images and the original images are decomposed into a serial of subbands by Contourlet transform. The ratio of the distributed energies of corresponding subbands between watermark image and original image is regarded as self-adaptive embedded strength. The watermark image is embedded into original image to form the embedded image. The embedded image has invisibility for watermarks. A blind detection approach from the embedded image is presented. The approach needs the embedded image only, and realize genuine blind detection watermark. Experiments show that the approach has good extracting effect and robustness.