Comparison of Image Quality Assessment Metrics for Evaluation of Performance of Anisotropic Diffusion Filter for SAR Images

Abstract

The reduction of speckle noise is an important part of synthetic aperture radar image processing. In the process, it is necessary to select the values of some filter parameters that will give the highest processing quality. To select such parameters, many authors use image quality assessment metrics. The question, however, is which of the many available metrics is optimal for this task. The article presents a comparison of image quality assessment metrics applied to filtering speckle noise on SAR images. For this purpose, a specially designed test image and algorithm for finding optimal parameters of speckle noise reduction filters, developed by the authors, are used. The study is carried out using the anisotropic diffusion filter as an example.