Impact of Contextual Filtering on TomoSAR Detection

Multi-baseline multi-temporal Synthetic Aperture Radar (SAR) techniques such as SAR Tomography (TomoSAR) are affected by different forms and sources of noise. Its presence in complex-valued interferograms alters the reliability and accuracy of height estimation results. Thus, the key challenge of TomoSAR is to identify scatterers interfering within the same resolution cell. To this aim, we propose a spectral contextual filtering method based on subbands decomposition to reduce noise influence and improve the quality of the interferometric product for TomoSAR application. The impact of the proposed pre-processing approach on the Generalized Likelihood Ratio Test-based detection is carried out. Experimental results on data acquired by TerraSAR-X sensor show the performances of the denoising method to increase the detection capabilities.