Research The Vehicle Detection And Vehicle Type Recognition Based On Video Sequences

It is very important for the traffic control and management that is gathering traffic information in time. But existing traffic information collection methods can not satisfy the demand of traffic control and management. Recently, with the development of the computer technology, it is widely used in traffic information collection that is the technology of the video. Compared with others detection methods, it has several advantages,such as wider detection scopes, more flexible setup, easier adjustment, convenient installation and maintenance, richer traffic flow parameters. Therefore it is significance for detecting type parameters of moving vehicles by researching the crucial video technology.To recognize the vehicles accurately and speedy, this paper presents a method based on Harris corner detection. The step is following:(1) This method obtains the accurate Harris corner of moving vehicles based on background subtract, and uses the iteration method to update background, improves background differencing and the optimum threshold by the Otsu's, so the moving object can be extracted accurately.(2) It selects three Harris corners to denote three type of vehicles which are landaulet, in-between car and large car. After Denoise calculates the average distance of Hausdorff between the Harris corner and the three types of swatches which need to be recognized. The two whose average distance of Hausdorff is smallest can be judged as the same type.Based on the theory studying, this paper give the identified system of vehicle type by Visual C# which bases on the operated system of Windows. Experiment results show that the method can give a high recognition precision with a better time-consuming.