The bootstrap Kernel-diffeomorphism filter for satellite image restoration

Abstract

The satellite imagery is very important in several fields such as security, agriculture and other fields. As like as other images, satellite images are subject to be degraded due to noise effects that occur during the capture and/or transmitting process. These effects will cause altered noise styles such as, speckle noise, Gaussian noise and others. The main purpose of the image restoration process is to eliminate the noise that present in the image. Researchers used linear and nonlinear filters to recover images. The Kalman linear filter is generally used. Non-linear filters like the extended Kalman filter (EKF) was also used. Bootstrap method is based on both Bayesian state estimation and Monte Carlo method, and it is a robust method because it is not constrained by the linearity in linear model presumptions. In this paper, we propose a Bootstrap kernel-diffeomorphism filter (BKDF) to reduce speckle noise in satellite images. We evaluated the performance of the BKDF by comparing it with the EKF according to the numeric values based on the image signal to noise ratio (ISNR) and peak signal to noise ratio (PSNR). Our results declare that BKDF has more efficiency than the EKF in the satellite image restoration.