The detecting and tracking system for vehicles

Abstract

It's very important to monitor the surroundings while driving the car. This paper uses camera to detect the location of vehicles and that assists drivers in driving more safely through the alerting system. Meanwhile, there were two different algorithms tested in the vehicle detection system, including Haar with Adaboost, and HOG-PCA with SVM methods. The preliminary results of this paper have been able to successfully detect vehicle and track vehicle location. In the vehicle detection, the average precision rate of 39.77% has raised to 82.45%. The vehicle tracking section has presented an improved version of the particle filter. Furthermore, take 200 particles for the test, vehicles could be tracked accurately in and out of the tunnel in the dramatic changes of light.