SAR image denoising based on patch ordering in nonsubsample shearlet domain

Abstract

Synthetic aperture radar (SAR) has been extensively adopted in a variety of fields, e.g., agriculture andmarine fields. In this regard, the improvement of SAR image quality has aroused a wide concern worldwide. In recentyears, image processing based on local patches has been very popular and proven feasible. In this paper, a novelSAR image denoising algorithm is proposed in the NSST domain on the basis of patch ordering. First, the shearlettransform is applied to logarithmic transformation of the noisy SAR image. Second, the coefficients of the shearlet aredenoised respectively by combining patch ordering and 1D filtering. Finally, the denoised SAR image can be obtainedby exponential transformation after applying the inverse shearlet to denoised coefficients. The experimental results showthat the proposed method not only effectively suppresses the speckle noise and improves the PSNR and ENL of denoisingthe SAR images but also obviously improves the visual effects of the SAR images, especially in maintaining the imageedge and texture information.