Research On Recognition And Location Of Moving Vehicles In Complex Environment

The further study for vehicle distance measurement,which is one of the key technologies of intelligent transportation system,will be valuable to theory and application.The technology based on stereo vision is becoming the main technical means of vehicle distance measurement as its advantages of non-contact,high accuracy and so on.The current dynamic vehicle measurement technology is easily affected by the environment and other external conditions,especially complex traffic scenes or inclement weather environment.Therefore,a robust vehicle distance detection model based on the principles of human memory mechanism is established in this paper.Moreover,the paper tries to build a vehicle distance measurement model based on the improved bat algorithm.The detailed work of thesis is as follows:(1)The Memory-based Vehicle Recognition Model(MVRM)based on human memory mechanism is built for dynamic vehicle recognition.In order to realize the accurate identification of the moving vehicle based on eagle eye's visual attention model,the model has been improved in two aspects.On the one hand,the rare-motion feature is introduced to measure the visual saliency,which improves the accuracy of the visual attention mechanism.On the other hand,according to the cognitive process of human visual information processing based on human memory mechanism,a model of vehicle salient region recognition is established.In this model,the instantaneous memory space and long term memory space are used to guide the process of vehicle target recognition,which improves the accuracy and robustness of the vehicle recognition.(2)In order to improve the insufficient precision of the traditional local matching algorithm,an adaptive weight stereo matching algorithm based on the improved bidirectional bat algorithm is proposed.Firstly,the initial matching result is obtained by the stereo matching algorithm based on adaptive weight,and the initial parallax provides timely obstacle avoidance information for vehicles obtained by the initial results.Then,the initial matching result is used as the initial value to optimize the matching points of the image based on the improved bidirectional bat algorithm.Finally,the accurate distance is obtained based on the information of the moving vehicle.The results of simulation experiment show that the dynamic vehicle recognition rate of MVRM model is 77.10%and the false recognition rate is only about 4.5%.There is good recognition effect in different kinds of weather conditions and complex road traffic environment.Simultaneous,the maximum error percentage of vehicle distance calculation is within 5%,and the computing time of each frame is 17.29ms.Experiments show that the algorithm proposed in this paper has good robustness and real-time performance,which can meet the needs of practical applications,and it is effective for the vehicle distance detection.