Method for detecting vehicle flow of multiple lanes on basis of video analysis

Abstract

The invention relates to a method for detecting vehicle flow of multiple lanes on the basis of video analysis, and belongs to the field of acquisition of traffic parameters of intelligent traffic systems. The technical scheme includes that the method for detecting the vehicle flow of the multiple lanes on the basis of video analysis includes steps of building video background models; setting transverse main virtual detection lines, auxiliary virtual detection lines and longitudinal virtual detection lines in video backgrounds; preprocessing video; building background image models by the aid of Gaussian mixture background models and detecting vehicle targets; determining whether vehicles need to be counted or not, in other words, determining whether change of relations between the vehicle targets is available in image data of each local frame on the corresponding main virtual detection line and image data of a next frame or not, according to the background image models of front and rear frames when the moving vehicles enter the main virtual detection lines; carrying out statistics on vehicle counting according to vehicle target detection results. The transverse main virtual detection lines and the auxiliary virtual detection lines are positioned on the lanes, the longitudinal virtual detection lines are positioned at the edges of the lanes, and parallel distances from the main virtual detection lines to the auxiliary virtual detection lines are fixed.

Description

technical field [0001] The invention relates to a multi-lane traffic flow detection method based on video analysis, which belongs to the field of traffic parameter collection of intelligent traffic systems. Background technique [0002] With the substantial increase in car ownership, it is very important to make high-quality use of the existing limited traffic resources to improve the automation level of urban traffic monitoring and command. To solve this problem, the research on Intelligent Transportation System (ITS) has been mentioned to a more important position. Among them, traffic flow detection is one of the important technologies and information that intelligent transportation systems need to consider. It provides an important data source for intelligent control. As the basic part of ITS, the traffic flow detection system occupies a very important position in ITS. It provides an important basis for subsequent traffic decisions such as vehicle speed measurement, lan...

AI Technical Summary

Problems solved by technology

[0005] Existing technologies are mainly aimed at expressway, expressway, weather environment, and road environment with good and clean vehicle detection, and cannot quickly adapt to scene changes and maintain a high detection accuracy rate under conditions such as rapid changes in weather and lane environments, etc. ; Existing technology cannot accurately detect traffic flow at intersections with dense vehicles, when vehicles press lines, change lanes, etc.

Images