Development of automatic target despeckling and detection methods in sar imagery

Abstract

Synthetic Aperture Radar (SAR) images play a significant role in newlinedifferent application fields like airborne, and civilian, and to observe various newlinescenarios over the horizon. SAR can work all day and in all weather conditions. newlineSAR is capable of penetrating through soil and cloud and plays a major role in newlineremote sensing and its applications. Modern SAR systems are capable of newlineproducing good-quality images. In most image processing systems, technical newlinebarriers are caused due to the noise characteristics in the background newlineenvironment. These technical barriers affect the ideal performance rates in the newlinesystem. SAR images are profoundly affected by speckle noise. Speckle noise newlineis multiplicative and it affects the image interpretations analysis. Therefore, it newlineis important to remove speckle noise. newlineThe estimation of noise is done by various noise reduction and newlineimage enhancement operations by using a Hybrid Laplacian Gaussian Filter newline(HLGF). This approach is to overcome the speckle noise, which has been dealt newlinewith in the first stage of the research work. Then, in the second stage of the newlineresearch work label, dependencies are modeled by using a probabilistic newlineapproach, Markov Random Fields (MRF) Ayed model, and optimal labeling is newlinedetermined by Bayesian estimation, in particular, Maximum a Posteriori newline(MAP) estimation. The main advantage of MRF models is that prior newlineinformation can be imposed locally through clique potentials. The primary goal newlineis to demonstrate the necessary steps to construct an easily applicable MRF newlineSAR image segmentation model and further develop its multi-scale and newlinehierarchical implementations. Moreover, the objective of this study involves newlinethe combination of these two methods in a multilayer model. newline