基于机器视觉的高速公路服务区违法上下客识别应用研究

Application Research of Illegal Boarding and Alighting Recognition in Expressway Service Area based on Machine Vision

为加强管控高速公路服务区违法上下客行为,提升执法效率,保障交通安全。本文基于违法上下客行为伴随的携带行李箱、背包等携物特征,利用高速公路服务区已有视频监控,基于检测区域的目标车辆识别与跟踪,采用Codcbook算法提取目标行人轮廓,通过人体骨骼关键点检测算法实现行人携物检测,结合行人移动方向与车辆停靠位置的相对分析,判别违法上下客行为。开发一套基于机器视觉的违法上下客识别原型系统,在试点服务区得到有效应用,对提升执法稽查效率及非现场执法智能化水平具有重要意义。

In order to strengthen the control of illegal boarding and alighting in expressway service areas,improve law enforcement effi ciency and ensure traffi c safety,this paper,based on the characteristics of carrying luggage,backpacks and other objects accompanied by illegal boarding and alighting behavior,uses the existing video monitoring in the expressway service area and the target vehicle identifi cation and tracking in the detection area.The contour of the target pedestrian is extracted by codcbook algorithm,and the pedestrian carrying detection is realized by human bone key point detection algorithm.Combined with the relative analysis of pedestrian moving direction and vehicle parking position,the illegal boarding and alighting behavior is distinguished.The development of a prototype system for illegal boarding and alighting identifi cation based on machine vision has been eff ectively applied in the pilot service area,which is of great signifi cance to improve the effi ciency of law enforcement inspection and the intelligent level of off-site law enforcement.