The Intelligent Tunnel Traffic Detection System Based On Video Processing

It is particularly important to ensure the safety of the tunnel traffic due to the closure of the tunnel structure. Tunnel real-time traffic safety monitoring system is an important means to ensure the tunnel security. In recent years, along with the development of intelligent transportation systems, video detection technology has been unprecedented developed. The smart tunnel traffic safety monitoring system based on video detection technology has incomparable advantages compared to traditional tunnel safety monitoring system. Video detection technology has an important significance to be applied to tunnel traffic safety monitoring.Moving target detection, localization and tracking are the key technologies of the video detection. This paper focused on theoretical analysis and experimental studies on the several key technologies.Then,the smart tunnel traffic safety monitoring system based on video-based detection technology has been developmented. The main works of the paper are listed as follows:(1) The choosing of moving target detection and vehicle location algorithm and experimental studies. The target detection algorithm based on the accumulated difference in depth is selected for the tunnel application after comparing several moving target detection and localization algorithms for the tunnel complex environment application. A differences in depth matrix is introduced to record the changes in the image in the modeling process of background, which can distinguish between the effects of noise and moving targets, and then we can basically eliminate the effect of noise. The results of experiment show that the background established by the accumulation of differences in depth background modeling method is clean, accurate, and strong anti-jamming capability.We can get accurate results using target detection algorithm based on the accumulation of differences in depth available. Watershed segmentation algorithm can get the contour of the region, and has small computational burden, high accuracy of segmentation. The watershed algorithm is selected for vehicle location.(2) Moving target tracking problem is a typical state estimation problem of dynamic system.The Kalman filter is the optimal solution of the tracking problem when the model of system is Gaussian linear. The Kalman filter tracking algorithm based on two level characteristics match has been proposed in this paper after the comprehensive analysis of existing moving target tracking algorithms. The two level characteristic match means the geometric characteristics of the target area match and nuclear histogram match. The moving target feature matching process is divided into two steps by the two level characteristic match.First,the geometric feature matching is finished, if the best match can be finded,then there is no need to do next level matching; or else, there goes the nuclear histogram matching. So we can ensure the accuracy of the tracking results while reducing the computational load. The results of experiment show that this algorithm can effectively track multiple moving targets, and has less computation, good real-time feature.(3) Finally, the intelligent transportation safety monitoring system based on video detection technology has been designed and implemented. The system consists of the following modules:image acquisition module, video storage and database modules, video and image processing module, the information management system module and other auxiliary modules. The results of experiment show that the system can detect traffic parameters and traffic incident of traffic flow,such as low speed, road share, speeding, illegal lane change and reverse drive detection. It meets the design requirements.