A Collaborative Despeckling Method for SAR Images Based on Texture Classification

Speckle is an unavoidable noise-like phenomenon in Synthetic Aperture Radar (SAR) imaging. In order to remove speckle, many despeckling methods have been proposed during the past three decades, including spatial-based methods, transform domain-based and non-local filtering methods. However, SAR images usually contain different types of regions, homogeneous heterogeneous regions. Some filters could despeckle effectively regions but not preserve structures well do suppress speckle effectively. Following this theory, we design a combination two state-of-the-art tools that can overcome their respective shortcomings. select best filter output for each area image, clustering Gray Level Co-Occurrence Matrices (GLCM) are used image classification weighting, respectively. Clustering GLCM use co-registered optical because structure information consistent, much cleaner than images. The experimental results on synthetic real-world show our method provide better objective performance index under strong noise level. Subjective visual inspection demonstrates has great potential preserving structural details suppressing noise.