An Iterative Semi Blind Transform Domain Watermarking

Abstract

Digital Watermarking is an efficient way of securing various multimedia contents. An Iterative semi-blind watermarking technique is proposed in this research for gray scale images, based on Transform Domain and Independent Component Analysis (ICA) Algorithms. Transform Domain includes Discrete Cosine Transform (DCT) and Singular Value Decomposition (SVD). In embedding process, the watermark is embedded into singular values of the host image's DCT coefficients. The host image and the secret image are considered as two separate signals and combined using the Fast-ICA Algorithm, a variant of Independent Component Analysis (ICA) mixing model, to generate the watermarked image and is also used to extract the watermark efficiently. The statistical property of the Fast-ICA Algorithm extracts the watermark differently at different time and hence the watermark is extracted repeatedly until the Normalized Cross Correlation (NCC) is above a certain threshold value to extract a clear watermark. The metrics Peak Signal to Noise Ratio (PSNR) and NCC are computed to measure the quality of watermarked image and extracted watermark respectively. This method increases the capacity, robustness and security than the Discrete Wavelet Transform (DWT) combined with SVD and Fast-ICA where the watermark is embedded and extracted from the low frequency band.