In this project, three essential characteristics of the HVS model namely – luminance sensitivity, edge sensitivity computed using block threshold and contrast sensitivity computed using block variance are employed to obtain the output (weighting factor) of a Fuzzy Inference System (FIS). The weighting factor is used to robustly embed a randomly generated watermark normalized as per N(0, 1) with mean 0 and variance 1 in the low frequency DCT coefficients of a given host image according to a pre-specified formula. The watermarked image has good perceptible quality and the computed values of MSE and PSNR indicate a robust embedding process.

2 3 4 5