Evaluating the Impact of DFT based Speckle Reduction Filter on T3 Matrix Elements in Polarimetric SAR Imagery

Abstract This paper presents a comprehensive evaluation of a novel speckle reduction framework tailored for polarimetric synthetic aperture radar (PolSAR) data. The study investigates the framework's performance across various polarization data configurations, including Full Polarimetry (Full Pol) and Single Polarization (Single Pol), operating across different frequency bands such as L, and S band. With a specific focus on T3 matrix elements (T11, T12, T13, T22, T33, and T23) and full polarimetric (full pol) data, the primary objective is to conduct an in-depth comparative analysis of the proposed framework against established speckle reduction techniques. To achieve this, the proposed framework is rigorously compared with well-known methods such as Refined Lee, Lopez, Lee Sigma, Non-Local Means, and IDAN filters across the aforementioned polarization and frequency configurations. Performance evaluation is conducted using a diverse set of metrics, including Signal-to-Noise Ratio (SNR), Equivalent Number of Looks (ENL), Coefficient of Variation (CV), Speckle Suppression Index (SSI), Speckle Suppression and Mean Preservation Index (SMPI), and Structural Similarity Index Measure (SSIM). The results of this evaluation demonstrate the superior performance of the proposed framework, with significant improvements in speckle reduction and preservation of image detail compared to existing techniques. These findings highlight the potential of the proposed framework as a valuable tool for the processing and analysis of PolSAR data across different polarization configurations and frequency bands, providing a comprehensive solution to the challenges posed by speckle noise in SAR imagery.