A new method for denoising SONAR images
Alexandru Isar, Sorin Moga, Dorina Isar · 2006
The SONAR images are perturbed by a multiplicative noise called speckle, due to the coherent nature of the scattering phenomenon. The use of speckle reduction filters is necessary to optimize the images exploitation procedures. This paper presents a new speckle reduction method in the wavelets domain using a novel Bayesian-based algorithm, which tends to reduce the speckle, preserving the structural features (like the discontinuities) and textural information of the scene. First, the different wavelet transforms are investigated and arguments to select the dual tree complex wavelet transform are presented. Next, accurate models for the subband decompositions of SONAR images, that permit the construction of maximum a posteriori filters, with closed-form input-output relations, are investigated. A blind speckle-suppression method that performs a nonlinear operation on the data is obtained. Finally, some simulation examples prove the performances of the proposed denoising method. These performances are compared with the results obtained applying state-of-the-art speckle reduction techniques.