Study Of Vehicle Detection And Counting In ITS

with the quick increase of economy, the amount of vehicle becomes larger and larger, How to make improvement on the status of traffic becomes the focus that the international governments and the departments pay attention to. As a result, The technology of ITS developed quickly. Every country has devoted great energy to study it. On account of the virtue of ITS based on video, it becomes the hotspot of studying. The technology of vehicle detecting is the foundation of ITS based on video. It plays a great role on traffic surveillance and controlling, and has great impact on work after them directly. This thesis studies several difficult problems in the process of extracting moving object.1. This thesis adopts the background suppression based RGB color to abstract vehicle area. We acquire the picture of background by background reconstruction based on T distribute, then update the background at the same time when abstract vehicle, as a result, the background we acquire can reflect the change of environment, ensuring vehicle abstracting accurately. Results show that this method can extract more integrated area contrast to other method. Besides, this thesis takes account of the case that the gray-level of environment changes rapidly, and we add the value of change to the threshold in order to extract the accurate vehicle areas.2. The area extracted contains shadow, it makes the area of vehicle becomes larger, and the vehicles near will be connected together, This thesis studies two methods of shadow detect, one is shadow detecting based on HSV color model and the other is shadow detect based on the information of blue band information.3. Extracting every vehicle accurately is a difficult problem in the case of occlusion and conglutination, this thesis attempts two methods to solve it, one is fuzzy K-means clustering, this method can give the number of vehicles in the area which exists occlusion and conglutination, but can't separate different vehicles;and the other method, this thesis gives the principle of determining the area existing occlusion toward one case of occlusion and conglutination usually (defined left-right occlusion), separating different vehicles by analyzing the shape of the area. This method has solved the occlusion problem partly.After the above operations, this thesis gives the instantaneous number of vehicles, this number can reflect real-time traffic status compared with the number of vehicles passed in unit time, it can guide traffic controlling.