Hybrid semi-blind image watermarking in redundant wavelet domain

Image watermarking in wavelet domain has been found useful for copyright protection and rightful ownership. Classical wavelet transforms, like discrete wavelet transform (DWT), are shift sensitive and provide information in horizontal, vertical and diagonal directions only. Shift invariance and directional information are required for better reconstruction of images. In this work, we propose a semi-blind gray scale image watermarking technique in redundant wavelet domain. The primary focus of this research is to highlight the usefulness of redundant wavelet transforms in image watermarking. We have used nonsubsampled contourlet transform (NSCT) and redundant discrete wavelet transform (RDWT). These redundant transforms are shift invariant. Also, NSCT provides rich directional information. Thus, they overcome the shortcomings of DWT and are more useful in image watermarking. We have integrated singular value decomposition (SVD) in the proposed method. We use NSCT-RDWT-SVD decomposition for single and dual image watermarking. For single watermark embedding, cover images are sub-sampled followed by one level NSCT and RDWT decomposition. SVD has been applied on obtained RDWT coefficients. In the same way, image watermark has been processed and NSCT-RDWT-SVD decomposition has been applied on it. SVD coefficients of the cover image and watermark image have been combined using scaling factor. Inverse SVD-RDWT-NSCT operation together with reverse sub-sampling provides watermarked image. For extraction of the watermark, we follow the NSCT-RDWT-SVD decomposition and SVD coefficients have been separated by the same scaling factor that was used in embedding. In the dual watermarking, Arnold transform has been used for encryption of the text watermark and rest of the steps are similar to single watermarking. NSCT, RDWT and SVD improve the performance of the proposed method against geometrical and image processing attacks for single and dual image watermarking. Experiments have been carried over standard images and results have been shown for natural and medical images. Qualitative and quantitative evaluations in terms of peak signal to noise ratio (PSNR), correlation coefficient (CC), bit error rate (BER) and structural similarity index metric (SSIM) show that the proposed method is suitable for single and dual image watermarking, and outperforms existing methods.