Clustering-Based 3-D-MAP Despeckling of SAR Images Using Sparse Wavelet Representation

Abstract

Image denoising is considered an effective initial processing step in different imaging applications. Over the years, numerous studies have been performed in filtering for different kinds of noises. The block matching with 3-D group filtering has added a new dimension and better results for denoising techniques. This work aims to establish a novel denoising method for multiplicative (speckle) noise employing 3-D arrays resulted from gathering similar patches in clustered areas of an image through the sparse representation based on discrete wavelet transform (DWT) and maximum <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a posteriori</i> (MAP) estimator technique. Experimental results justified a good quality of the filtered image by the novel framework, which appears to demonstrate better denoising performance against state-of-the-art algorithms according to the objective criteria [peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and edge preservation index (EPI)] values and subjective visual perception.