A New Object Detection Method Based on T Mutation PSO for SVM Parameter Optimization

In order to detect objects efficiently and accurately, we propose a high real-time and robust performance object detection method in this paper. Firstly, a feature called Locally Assembled Binary Haar-like (LABH) is selected as the feature vector and SVM is used for classifier. In addition, a new type of PSO called T mutation PSO (T-PSO) is used to select suitable parameters of SVM, which could improve the classification ability of SVM prominently. T-PSO is a novel optimization algorithm, which not only has strong global search capability but also helps to find the optimum quickly by using T mutation. The experiment carried out on videos from INRIA dataset, MIT dataset and Daimler dataset illustrates that the proposed method is real-time and feasible enough for the task of object detection.