Road type recognition based on SOM and SVM

It has the great significance to recognition road type based on vehicle load. In this paper, it is proposed that Self-Organizing feature Map (SOM) network is used to determine sample IDentity (ID) and Support Vector Machine (SVM) is employed to recognize road types based on vehicle load. Experiment results indicate that the improved Particle Swarm Optimization (PSO) algorithm gives best performance among all used methods in parameter optimization of SVM and Radial Basis Function (RBF) is the ideal kernel function of SVM. By cross-validation, the highest average recognition rate is 72.5%. This paper present a useful research for road type recognition based on dynamic load.