Stduy On A Method Of Vehicle-logo Recognition

With the increasing social wealth and the development of scientific technology,people's living standards greatly raise while social unrest and crime also increase. SoIntelligent Home has gained more and more attention. Researching on Vehiclerecognition which is a critical part of the Intelligent Home system has important valuein use.In this paper, attention is paid to the research on vehicle-logo recognition, and afast method to vehicle-logo location from coarseness to fine is proposed, which isbased on the prior knowledge of the license plate's location.For the low speed of therecognition system by the moment invariants that based on gray-level image can notmeet the demand of real-time, a new algorithm of vehicle-logo recognition is presentedin this paper. In this algorithm,the logo region is extracted as shape features by using ofmulti-threshold OTSU binarization, then the recognition is achieved based on momentinvariants of the logo-region and the minimum distance classifier. Experiments showthat the recognition algorithm has faster rate in the processing.The simulative test results show that the approaches to vehicle-logo location andvehicle-logo recognition have good precision and high speed.The average recognitionprocessing time is 1.2 seconds,and the recognition accuracy is 92%.