A new encrypted image watermarking based on DTCWT and random pixel exchange

Abstract

An encrypted image watermarking algorithm based on random pixel exchange and the alliance of SVD (singular value decomposition) and DTCWT (dual-tree complex wavelet transform) is proposed. Better directionality and shift invariance properties are motivated towards the selection of DTCWT. The invariance of SVD guarantees robustness and stability. The watermark image is preprocessed using random pixel exchange with a logistic key to get the encrypted version. Initially, the original image decomposed with DTCWT and selected the sub band based on the intensity invariance value. Applied the SVD on both selected sub band and encrypted watermark to generate the singular values. Next, the resultant singular values are combined with an appropriate scaling factor. To generate the watermarked image, performed the inverse DTCWT on modified sub bands. In the process of extraction, used the same scaling factor to generate the decrypted watermark image. With the correct logistic key, the watermark image is recovered. Further, transparency analysis and robustness analysis are carried out. High values of PSNR and normalized correlation (NC) indicates the better robustness and transparency of the proposed algorithm.