Enhancement of SAR images using nonlinear filters

Abstract

Synthetic Aperture Radar (SAR) imaging is used in various applications such as urban planning, land cover mapping and cartography. SAR imaging systems are very useful in acquiring information under different climatic conditions. Naturally, SAR images are disturbed from speckle noise, which degrades the performance of automatic SAR image interpretation and analysis. Hence, the SAR image analysis system starts with a preprocessing step to remove speckle noises before preceding major steps like segmentation and classification. This research work proposed three filtering methods for despeckling the SAR images.The main objective of this research is to design efficient filters for despeckling by avoiding the usage of fixed window size as well as preserving other image characteristics. The proposed filters are: (i) Hybrid Filtering, (ii) Intensity Coherence Vector (ICV) based Filtering and (iii) ICV based Non-Local Means Filtering. The hybrid filtering takes advantage of all the filters while the demerits of each filter could be resolved by merits of other. For hybridizat newline