Improved robust digital image watermarking with SVD and hybrid transform

In this paper two methods of color image watermarking are proposed. Both methods use SVD and DCT-Walsh hybrid transform. First method uses SVD on column hybrid transformed host and watermark images. Second method uses SVD on sorted column hybrid transform coefficients of host and watermark images. In both methods, low frequency transform coefficients are used in contrast to traditional way of using middle frequency coefficients. In first method, 30 singular values of watermark are needed per plane to embed it in host. In second method, this number is drastically reduced to 3 values. This is possible because high energy compaction property of SVD gets further enhanced due to sorting of transform coefficients prior to SVD. Performance of both the methods is compared for robustness against various compression, noise addition, resizing, cropping, and histogram equalization.