Vehicle Type and Sub-type Recognition by Lamp Pairs Distance and Lamp Contour in ITS

Abstract

For vehicle surveillance and communication control in ITS, it's very important to recognize the vehicle type and sub-type. The algorithm based on lamp pairs distance and lamp shape is proposed in this paper. In this method the distribution of vehicle lamps are retrieved in preprocessing, then lamp type table (LTT) is calculated according to the lamp pairs distance. Because the lamp pairs and the lamp contour are intrinsic attributions of every type of automobiles, this algorithm uses these composite features to recognize the vehicle type and sub-type. We use the associated light pairs distance and contour information to recognize the vehicle type and sub-type. Experiments verified the satisfied effect of this algorithm. In the experiments a huge amount of vehicle surveillance images are utilized to recognize vehicle type and sub-type and test the accuracy of recognition.