Zero-watermarking Algorithm for Color Image Based on DWT-DCT-SVD

Aiming at the conflicts between invisibility and robustness for the digital watermarking algorithm by modified the data of the original image, a zero-watermark algorithm is proposed for color image based on DWT, DCT and SVD. First, the method is to switched the host image from RGB color space to YCbCr space. Then, by using a layer of discrete wavelet transforms to the luminance Y of the color image, and divided the low frequency band into blocks by using discrete cosine transform, and conducted SVD with every block to generated the zero-watermark by the decomposition of the singular values as the characteristics of the image. Finally, the conflicts between invisibility and robustness of the watermarking is better solved to combined with the idea of zero-watermark by the strong stability of singular values about image to resist geometric distortion. And the process of extracted the embedded watermark without the original image to achieved a blind detection. Experiments show that the proposed algorithm has strong robustness to against familiar signal processing method, such as cuting, median filtering, salt and pepper noise and JPEG compression.