A Study of Multiplicative Watermark Detection in the Contourlet Domain Using Alpha-Stable Distributions
The contourlet transform can effectively provide sparse and decorrelated image representation. And its subband coefficients can be modeled as the generalized Gaussian (GG) distribution. In this paper, an improved maximum likelihood (ML) parameter estimation method is proposed, in which a novel initial estimation value and a modified iterative algorithm are used. The new approach has been applied to the contourlet-based texture image retrieval. Experimental results show that, compared with the current ML estimation method, the proposed approach can more accurately estimate the GG distribution parameters, and more effectively improve average retrieval rate on the VisTex database of 640 texture images.