Research And Implementation Of Ambulance Recognition In Traffic Video Image

The ambulance carrying hope of life,every minute counts.However,in the actual road conditions,ambulances are often blocked in the middle of the road.The realization of intelligent transportation system can effectively relieve traffic congestion,improve the efficiency of transportation,and reduce the occurrence of traffic accidents.Intelligent Traffic Systems(ITS)is the predecessor of the Intelligent Vehicle Highway System(IVHS).ITS combines advanced communication technology,sensor,electronic control technology and digital image processing technology,so as to realize intelligent management of transportation system,Set up a wide coverage,full-featured and real-time,accurate,efficient transportation management system.The recognition of the ambulance is the use of video and digital image processing technology to achieve the classification of this particular vehicle.Thus,it is possible to record and inquire the location information and driving information of the ambulance.Provide the basis for traffic police command and dispatch vehicles and other special operations.The innovation of this paper is embodied in the following two aspects:First,the vehicle recognition method based on feature extraction and classifier training is researched in this paper.First,based on the feasibility analysis of the HOG characteristic to identify the ambulance vehicle model,this characteristic is selected as the identification feature of the ambulance vehicles.Second,using SVM as the classifier to carry on the classification training,which can realize the identification of the ambulance vehicles.Then,through the normalized product correlation matching and invariant moment matching methods to scan and match the vehicles that meet the ambulance model.Compared with the existing methods,the HOG algorithm is added to the white color feature to identify the ambulance model,so the accuracy of the vehicle identification is improved.Except that,in this method the matching accuracy of the Red Cross logo of ambulance is improved by two times matching.Finally,the effectiveness of the proposed algorithm is verified by experiments.Second,because the sparse data representation can approach the original data feature,So the use of sparse-representation to distinguish the target can reduce the amount of computation and processing costs,At the same time,in order to avoid a large number of sample images as a training object to bring a very complex calculation,In this paper,we use a small number of optimal basis matrix as the sample image features to train the dictionary.So in order to realize the identification of ambulance,a kernel K-SVD dictionary training algorithm based on signal representation is proposed to make the classification and identification of vehicles.And then do the color characteristics detection for the vehicle which have been classified to check if is consistent with the ambulance color characteristics.The experimental results show that this algorithm can improve the recognition efficiency.Finally,the simulation system is designed and implemented according to the recognition algorithm of this paper,and the system interface and function modules are introduced.