Alternative Methods Based on Subtraction and Linearization of the Noise Fields for Wide-Beam SAR Image Denoising

The two main sources of noise, thermal noise and the scalloping effect, could have a significant impact on the cross-polarized synthetic aperture radar (SAR) images in the extra-wide (EW) mode. Although the noise vectors are provided in annotated files of the data for the subtraction, they may not be sufficient to remove the entire noise effectively. This paper presents alternative methods based on the noise fields of the training area, which should be chosen from the scene. By linearizing the thermal noise of a chosen training area, the suggested approaches can reduce thermal noise. Alternatively, they can extract thermal noise from the training area and then modify it to make it appropriate for subtraction from the complete SAR image in the range direction. These techniques were used on two separate horizontal-vertical (HV) polarized SAR images, and they both performed well in terms of reducing thermal noise and enhancing the HV-polarized EW SAR images. Moreover, the scalloping effect could be lessened by eliminating the scalloping effect's slope.