Classification of Synthetic Aperture Radar Images using Particle Swarm Optimization Technique

Abstract

In this thesis, SAR image classification problem is considered as an optimization problem various clustering techniques are addressed in literature for SAR image classification. This thesis focuses on an evolutionary based stochastic optimization technique that is Particle Swarm Optimization (PSO) technique for classification of SAR images. This technique composes of three main processes: firstly, selecting training samples for every region in the SAR image. Secondly, training these samples using PSO, and obtain cluster center of every region. Finally, the classification of SAR image with respect to cluster center is obtained. To show the effectiveness of this approach, classified SAR images are obtained and compared with other clustering techniques such as K-means algorithm and Fuzzy C-means algorithm (FCM). The performance of PSO is found to be superior than other techniques in terms of classification accuracy and computational complexity. The result is validated with various SAR images.