| With the progress of science and technology,the increase of urban vehicles and people’s safety awareness,the construction of intelligent transportation has been more and more attention,and the intelligent transportation system has been widely used in the road monitoring.Currently vehicle license plate recognition technology is mainly used in the intelligent transportation system,however,the technology for "unknown license plate" vehicle such as no license and have been blocked,can not be effectively identified.The recognition and tracking of the "unknown license plate" vehicle is an urgent problem to be solved in the traffic law enforcement.This paper focuses on the consistency analysis of "unknown license plate" vehicle object on the highway.The vehicle consistency analysis is that when the target vehiclepass through a bayonet,the system find the target vehicle through complete consistency analysis by the method of pattern recognition in adjacent bayonet.The consistency analysis of a vehicle is divided into consistency analysis of vehicle objects in the same bayonet and in adj acent bayonet.This paper mainly focuses on the consistency analysis of vehicles in adjacent bayonet.In the consistency recognition,firstly the target vehicle must be detected and extracted from the decoded video.Secondly,feature recognition is done based on vehicle characteristics.Finally,we establish the consistency recognition model to analyze the consistency of vehicles.In the aspect of detection and extraction of the vehicle key frames,this paper analyzes the key frame extraction method based on the mask line width and profile center,which are easy to be affected by the vehicle in the adjacent lane when the key frame is extracted.Thus this paper propose a key frame extraction method based on the road boundary constraint which can effectively avoid the influence of the vehicle object in the adjacent lane.According to the color and geometric features,the system can identify the color,the vehicle type and the carriage type.In vehicle consistency recognition model,firstly,the color of the vehicle is confirmed by the identification of the color feature,and then the vehicle type and the carriage type are determined by the 3D geometric features.If these classification is same,characteristic similarity is calculated,and then we go further analysis and identification using scale invariant feature.Finally,we sum the similarity of each feature and obtain the overall similarity.The vehicle consistency recognition model of "unknown license plate" vehicle is established in this paper,which use the consistency recognition model to analyses the"unknown license plate" vehicle and has a better real-time and a more accurate analysis results,also can be effectively applied to the existing traffic.video monitoring system.Through the consistency recognition can accurately identify the similar vehicles of targets "unknown license plate" vehicle or target "unknown license plate" vehicle,which provides a feasible basis for the traffic police to the law enforcement of the"unknown license plate" vehicle,and provides a powerful technical supplement to the construction of the intelligent transportation system. |
