| Intelligent traffic management system is the future of traffic management in21th Century. With the rapid development of Expressway and vehicle management system, there is an opportunity of practical applications for intelligent traffic management system. A lot of traffic parameters ara got by the intelligent traffic management system, where the parameter of traffic flow is an important one. So the technology based on traffic video detection has been an important research field of intelligent transportation managment system.This paper provides a real-time detection method for road vehicles according to the actual traffic situation and images from it, aimed to detect the vehicle target quickly and accyrately from the complex background. After exploring and studying problems such as video image pre-processing, vehicle detection based on the summary and analysis of the existing vehicle detection method, we provide a combination algorithm using frame difference algorithm and Haar-like rectangle feature. The effectiveness of the new method are proved by experiment.This paper analyses the forms of fast calculation method with Haar-like rectangle feature and integral image on the base of existing research, and use the AdaBoost algorithm to selcect12rectangle features to represent characteristics of the vehicle. The shadow parts underneath vehicle is represented by single feature, and the two vertical edges and shadow part of vehicle combines an "U" shape feature.In this paper, according to the different requirements of vehicle recognition, realized the algorithm in the VC platform called the standard library function of OpenCV using C language on various requirements of detection. We applied the algorithm to the images got from the real-time road and tested:traffic flow statistics, and got the experiment data and performance evaluation under the diversity of shape and complexity of environment according to the content and results of it. |
PDF Full Text Request