Intersection lane division detailed traffic parameter acquisition method based on double cameras

Abstract

The invention discloses an intersection lane division detailed traffic parameter acquisition method based on double cameras. Due to the inherent defects of a single low-angle camera, two low-angle cameras with opposite visual angles are used for jointly monitoring the entrance road section of the intersection. Firstly, double cameras are installed at an entrance road section of an intersection, and offline setting of a region of interest is carried out; secondly, vehicle robust detection represented by a vehicle head and a vehicle tail is realized for each camera, and the position of each vehicle on each lane is divided by using stable features; after the exact time of the vehicle passing through the predetermined detection line and the parking line is further estimated, the real-time estimation of the cumulative curve of arrival and departure of the vehicle in different lanes can be realized. And finally, based on the cumulative input and output model, calculating traffic detail parameters such as the average arrival rate and the departure saturation flow rate of the divided lanes and the queuing length of the vehicles in the divided lanes by utilizing the arrival and departure cumulative curves of the vehicles in the divided lanes. The method has important significance in intelligent traffic monitoring and intersection traffic parameter acquisition.

Description

technical field [0001] The invention belongs to the technical field of intelligent traffic monitoring system and intersection traffic parameter acquisition. Using computer video intelligent processing technology, two low-angle cameras with opposite viewing angles are used to jointly monitor the entrance section of the intersection, and realize the detailed traffic parameter acquisition of lanes at the intersection. . Background technique [0002] Road intersections are an important part of urban road systems. Quantitative evaluation of intersection capacity, delay and service level is of great significance for the optimization of traffic organization at intersections and the timing and control management of signal lights. The detailed traffic parameters of the lanes, represented by the length of vehicle queuing, the arrival rate of vehicles at suitable positions upstream of the entrance section, and the saturation flow rate of vehicles leaving the stop line during the green...

AI Technical Summary

Problems solved by technology

[0004] However, there is a fundamental problem in the current queue length detection method based on image processing technology: the calculated queue length can only be expressed as the distance from the stop line at the entrance section (regardless of the pixel distance in image space or the distance in the three-dimensional real world ), rather than the count of vehicles that actually need to stop and wait

Most intersection low-angle video vehicle detectors are mainly installed near the stop line of the entrance section, and the installation height range is limited to within 6 meters, which can only realize the traditional vehicle queue length detection method measured by distance

Images