S.C. Satapathy et al. (Eds.): Proc. of Int. Conf. on Front. of Intell. Comput., AISC 199, pp. 747–754. DOI: 10.1007/978-3-642-35314-7_85 © Springer-Verlag Berlin Heidelberg 2013 Despeckling of SAR Images via an Improved Anisotropic Diffusion Algorithm Anurag Gupta, Anubhav Tripathi, and Vikrant Bhateja Deptt. of Electronics and Communication Engineering, Shri Ramswaroop Memorial Group of Professional Colleges, Faizabad Road, Lucknow-227105, (U.P.), India {anuraggpt8,proabhi9,bhateja.vikrant}@gmail.com Abstract. Synthetic Aperture Radar (SAR) is a powerful tool for producing high-resolution images but these images are highly contaminated with speckle noise. This paper proposes an improved Anisotropic Diffusion Algorithm for despeckling SAR images. The proposed algorithm is obtained by using a diffu- sion coefficient which consists of a combination of first and second order deriv- ative operators. The spatial variation of this diffusion coefficient occurs in such a way that it prefers forward diffusion to backward diffusion resulting in im- proved structural details and edge preservation. The simulation results also show better computational efficiency in comparison to other denoising techniques. Keywords: Speckle, SAR images, Diffusion coefficient, Multiplicative noise. 1 Introduction Speckle is a kind of multiplicative noise that affects most of the coherent imaging systems. The presence of speckle noise in an imaging system reduces its resolution; especially for low contrast images such as Synthetic Aperture Radar (SAR) images. This creates problem in automatic processing of SAR images, used in various applica- tions like crop monitoring, search and rescue operations, military target detection etc. Therefore, the suppression of speckle noise is an important consideration in the design of coherent imaging systems. Over the last few years, various despeckling techniques for SAR images have been proposed [1-3]. Among them, Anisotropic Diffusion (AD) filters [4] based on nonlinear heat diffusion equation surpass most others in terms of accuracy and robustness. These filters fall into the category of Partial Differential Equation (PDE) based image processing, originated from the work of Perona and Malik [5]. This method was capable of reducing the noise content of the image as well as enhancement of the boundary information within the data. However, this filter [5] introduced blocky effects and it blurs the edges with the number of iterations of the filter. Yu and Acton, therefore introduced an edge sensitive diffusion method, called Speckle Reducing Anisotropic Diffusion (SRAD) filter [6], which defined an instan- taneous coefficient of variation to detect the edges in the noisy images. Aja-Fernández and Alberola-López [7] further developed this method by introducing a new AD filter,