A Statistical Speckle Suppression Algorithm for Underwater Laser Image Based on Nonsubsampled Contourlet Transform

Underwater laser image is seriously contaminated by speckle noise which is due to the coherent nature of the scattering phenomenon. In this paper, the authors propose an adaptive speckle suppression algorithm via the novel nonsubsampled contourlet transform. The statistical model for speckle noise is first analyzed to obtain a simple and tractable solution in a closed analytical form. Gaussian distribution for speckle noise and a general Gaussian distribution are adopted to model the statistics of contourlet coefficients in logarithmically transformed laser images. Then based on the maximizing the a posteriori estimation with the assumption that speckle noise is spatially correlated within a small window, we utilize a locally adaptive Bayesian processor whose parameters are obtained from the neighboring coefficients in highpass subbands. Experimental results show that comparing with classical wavelet method, the proposed algorithm shows a superior performance in suppressing the speckle noise and retaining geometrical structures of the image.