Despeckling of synthetic aperture radar images using shearlet transform

Anotace: Synthetic Aperture Radar (SAR) is widely used for producing high quality imaging of Earth surface due to its capability of image acquisition in all-weather conditions. However, one limitation of SAR image is that image textures and fine details are usually contaminated with speckle noise. This noise is multiplicative and having granular pattern. It is caused due to the integration of various backscattered signals obtained during SAR image acquisition. It degrades the quality of SAR image and causes difficulty in visual interpretation. This paper presents a speckle reduction technique of SAR images based on statistical modelling of detail band shearlet coefficients (SC) in homomorphic environment. Modelling of SC obtained from log transformed noiseless SAR image are carried out as Normal Inverse Gaussian (NIG) distribution. However, SC pertaining to speckle noise are modelled as Gaussian distribution. These SC are segmented as heterogeneous, strongly heterogeneous and homogeneous regions depending upon the local statistics of SAR image. Then maximum a posteriori (MAP) estimation is employed over the SC of all regions of detail bands that belong to homogenous and heterogenous category. Objective and subjective quality assessment is performed for synthetic images and real SAR images. Reference image quality metrics, PSNR and SSIM for synthetic images and non-reference image quality metrics, ENL for real SAR images signify the potential of proposed method in comparison to six existing image denoising methods.