SAR images denoising using a novel stochastic diffusion wavelet scheme

The rapid processing of remote sensing (RS) images is crucial in real-time monitoring. However, the computation cost of RS is high and traditional methods are not effective. Cloud computing with its capability of parallel computing provides an effective service for executing RS processing. A cloud-integrated web platform for maintaining geographic information system (GIS) and RS application such as oil spill detection, meteorological monitoring through synthetic aperture radar (SAR) images is a fast-growing application. SAR RS helps also in studying land and sea-based phenomena, especially the capability of acquiring weather forecasts all-day. Wavelet transform is a very well-known tool for prime applications in time series, function estimation, and image analysis. In this work, an effective non-deterministic polynomial computation technique for noise mitigation of SAR Images which has its basis on Hybrid wavelet transform (WT) is proposed. Proper noise reduction parameters should be chosen while selecting wavelet transform for SAR images. So, in the proposed method Stochastic diffusion search (SDS) optimization algorithm is utilized for selecting the optimal noise reduction parameter thus leading better filtration performance. Effective choice of wavelet noise mitigation techniques includes wavelet function, decomposition levels, and threshold selection rules for improved noise reduction. Experimental results show that MSE, PSNR and standard deviation are enhanced with the proposed method.