Vehicle Type Identification Based on Car Tail Text Information

Based on feature matching of car tail text information, a novel approach for vehicle identification is proposed in this paper. Our method creatively implements Scale-Invariant Feature Transform (SIFT) to extract distinctive invariant features from car tail text sub-images, which is a new application of SIFT. In our approach, firstly, the coordinate and the content of a vehicle’s license plate are presented during the process of plate localization and recognition. Secondly, based on the plate location information, we apply a text information localization procedure which could be divided into two processes, robust localization and accurate localization. Then, a SIFT-based template matching method is provided to recognize the text information. Finally, we are able to determine whether the result conforms to the known vehicle type captured according to the plate license contents in the vehicles information file. The experimental results show a high recognition rate in acceptable time and prove the availability of vehicle type identification. Keywords-Scale-Invariant Feature Transform; vehicle type identification; car tail text information; image matching