Robust image watermarking using dual tree complex wavelet transform based on Human Visual System

Abstract

In this paper, we proposed an image watermarking algorithm based on Human Visual System's perceptual model in the dual tree complex wavelet domain. We took advantages of the regular and the complex wavelets (perfect reconstruction, approximate shift invariance, and good directional selectivity) for watermark embedding. Firstly we applied the JND (Just Noticeable Difference) to control the watermark embedding strength. Then the host image was decomposed into six sub-bands by the 4-level dual tree complex wavelet transform. We embedded the watermark sequence into the six subbands with the same JND value. Lastly the watermark was extracted by comparing the host image coefficients and the watermarked image coefficients. Experimental results confirm that the proposed method was robust against geometric distortions, Gaussian noise, JPEG compression and Gaussian low pass filtering attacks.