Effective Transform Domain Denoising of Oceanographic SAR Images for Improved Target Characterization

Synthetic Aperture Radar (SAR) images are widely used for a variety of applications such as surveillance, agricultural assessment and classification, planetary and celestial investigations, geology and mining, etc., due to its remarkable characteristic of capturing it under all weather conditions. SAR images are highly prone to speckle noise due to the ingrained nature of radar backscatter. Speckle removal is highly essential to limit the difficulty encountered while processing the SAR images. An exhaustive work has been done by researchers to despeckle SAR images using spatial filters, wavelet transform, and hybrid approaches. This work aims at exploring the different despeckling techniques to identify the best and suitable methodology. On measuring the despeckling performance using Peak Signal-to-Noise Ratio, Edge Preservation Ratio, Speckle Suppression Index, Speckle Suppression and Mean Preservation Index, and Structural Similarity Index simultaneously for the various techniques experimented, ridgelet transform-based thresholding works well. It gives better results by applying ridgelet transform and processing the subbands with minimax thresholding. The type and characteristics of the scene imaged also influence the result.