Remote sensing image classification with parameter optimized Support Vector Machine based on evolutionary computation

Remote sensing image classification has been widely applied in many fields such as resource exploration, environmental monitoring and urban planning. Support Vector Machine (SVM) is adopted in our research, to classify two sets of SPOT-5 images of an urban area. In order to achieve high classification accuracies, the kernel function of the SVM classifier is selected beforehand. Furthermore, the kernel parameters are also optimized using different evolutionary computation techniques, including Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Comprehensive Learning Particle Swarm Optimization (CLPSO). The best classification scheme is determined based on comparative experiments, and the final classification results fully support the monitoring needs and aid in the formulation of urban expansion and land reclamations.