An Improved Kuan Algorithm for Despeckling of SAR Images
Aditi Sharma, Vikrant Bhateja, Abhishek Narayan Tripathi · Advances in intelligent systems and computing · 2016
Synthetic Aperture Radar (SAR) is an acquisition tool for coherent imagery used for meteorological and astronomical purposes. The speckle noise diminishes the information and image quality which evokes the necessity of pre-processing of SAR images. Kuan filter is a popular despeckling algorithm among the various local statistics filters. Kuan filter works efficiently within the homogenous regions of the SAR images while penalty is imposed by the edges. This paper presents an improved Kuan filter which combines the concept of gradient and conduction function for despeckling of SAR images. In this, the image is processed by classifying it into three regions i.e., homogenous, non-homogenous and isolated regions respectively; depending upon the estimated value of noise parameters. This approach thereby provides a simple solution to overcome blurring across the edges during despeckling. Simulation results make it evident that the proposed despeckling algorithm yields better results as compared to other primitive filters.