A method for people counting using feature fusion based on SVR with PSO optimization

摘要

For low density crowd, the statistical information of pixels and feature points can reflect the change of crowd density. Therefore, pixels and corners are fused in this paper, then, SVR is used to learn the corresponding relationship between feature and the number of people. While PSO is used to optimize the choice of parameters C and gamma in SVR. The experimental results show that the SVR optimized by PSO has better prediction accuracy.