Vision-based preceding vehicle detection and tracking

This paper presents a preceding vehicle detection and tracking system by using Support Vector Machine-Based Particle filtering (SVMPF). SVMPF integrates the Support Vector Machine (SVM) score with sampling weights. The sample weights, which are used to construct a probability distribution of samples, are measured by the SVM score. Once the vehicle is detected and tracked, it changes to SVM tracking mode which is simpler than the previous SVMPF mode. In the experiments, we demonstrate that our system can track the preceding vehicles under different whether conditions.