SONAR images restoration in the wavelets domain

摘要

SONAR images acquisition is inherently affected by the presence of a multiplicative noise called speckle, on one hand, and the loss of pixels, on the other hand. The goal of this paper is to propose an image restoration system capable of reducing the speckle noise without affecting the image features and recovering the lost pixels. The methods used for the denoising part belong to the wavelets based category. We will compare denoising methods that involve the use of the Dual Tree Complex Wavelet Transform and three variants of bishrink filter and we propose a diversified denoising method that takes into consideration the structural and textural features of the image under analysis. The recovery of lost pixels is done using an exemplar based algorithm.