基于概率算法自适应更新背景的运动车辆检测

交通流量检测是智能交通系统中的一个重要研究方向和热点问题,基于视频的车辆检测是交通流量采集分析的核心技术,它为交通流量参数的实时获取提供了可能。为实现在复杂交通视频场景中实时准确检测各类的运动车辆,在研究传统背景差分算法的缺点的工作基础上,提出一个自适应的贝叶斯概率背景检测算法,进而完成了较准确的运动车辆分类检测。实验结果表明该方法具有高效实时的特点,能够较准确地实现复杂交通路面的背景提取和运动车辆的检测,具有良好的鲁棒性。

Adaptive real-time vehicle detection based on Bayesian rule background model

Efficient detecting and tracking of vehicles is very important for collecting traffic flow information in intelligent transportation systems.Visual based motion analysis of vehicles is an active research topic of traffic flow estimation which involves detecting,tracking and recognizing from the surveillance image sequences or videos.Efficient and robust vehicle detecting and tracking under the real complex road scene is still a challenge task.This paper presents an efficient traffic flow detection method which firstly extracts foreground image using Bayesian classification algorithm,and then detects vehicle object with background subtraction.The method features low computational load,thus meets the real-time requirements in many practical applications.It tests this method with very low/high quality traffic surveillance videos and gets high detection accuracies.