A reliable and secure image watermarking algorithm using homomorphic transform in DWT domain

Abstract

Image watermarking is one of the reliable methods which easily retains integrity and confidentiality of images against illegitimate users thereby ensuring copyright protection. In this paper a new image watermarking technique is proposed which employs fusion of discrete wavelet transform (DWT), homomorphic transform (HT), singular value decomposition along with Arnold transform (AT). In this work first level DWT is performed on host image which results in various subbands (LL, HL, LH and HH). HL subband is chosen which is further transformed by HT to decompose it in illumination and reflectance components. The watermark is scrambled with AT in order to strengthen security towards malicious attacks. Scrambled watermark is now embedded into singular values of the reflectance component. As reflectance component contains prominent image features, hence embedding in this part provides better immunity against several attacks and imperceptibility is also attained as this component varies very rapidly. Thus, major aim for proposed watermarking technique is achieved. The proposed scheme is implemented with numerous wavelet families such as Haar, Coiflet, symlet and biorthogonal. The main idea for implementing with different wavelets is to judge which wavelet demonstrates good robustness against various attacks. Proposed scheme performance is also evaluated by examining comprehensively under several attacks such as rotation, sharpening. Results clearly prove high robustness for the proposed scheme. Superiority of proposed scheme is clearly indicated by comparing it with other commonly—used schemes as reported by other researchers. Benchmark attacks like “Checkmark” attacks are also verified for the proposed scheme.