基于车底阴影的智能车前方车辆检测算法
|
摘要
对前方车辆进行实时精确的检测是智能车系统中对车辆进行横向和纵向控制的基础。本文提出一种鲁棒的车辆检测方法,可适用于有车道线场景和无车道线场景车辆检测问题。车辆检测可以分为车辆假设产生和车辆验证2个阶段。在车辆假设产生阶段,文中提出一种车底阴影二值化方法,可以不受车道线有无影响,对光照强弱、建筑物或树的投影等抗干扰能力强。在车辆验证阶段,文中提出车辆粗验证和精确验证2个步骤。粗验证阶段采用车辆对称性,初步定位车辆位置,然后在精确定位阶段采用机器学习方法进行精确定位。
The real-time and accurate detection of forward vehicle is the basis of the lateral and longitudinal control in the intelligent vehicle system. A robust vehicle detection method is proposed here, which can be applied in all scenes with or without lane line. The vehicle detection can be divided into two stages: vehicle hypothesis generation and vehicle verification. In the first stage, a vehicle shadow binarization method is proposed. In the second stage, two steps are proposed: rough verification and precise verification. In the rough verification stage, the vehicle can be positioned using the vehicle symmetry, and in the precise positioning stage, the machine learning method is used for positioning.
The real-time and accurate detection of forward vehicle is the basis of the lateral and longitudinal control in the intelligent vehicle system. A robust vehicle detection method is proposed here, which can be applied in all scenes with or without lane line. The vehicle detection can be divided into two stages: vehicle hypothesis generation and vehicle verification. In the first stage, a vehicle shadow binarization method is proposed. In the second stage, two steps are proposed: rough verification and precise verification. In the rough verification stage, the vehicle can be positioned using the vehicle symmetry, and in the precise positioning stage, the machine learning method is used for positioning.
引文
[1]韵卓.基于DSP的道路前方车辆识别算法研究[D].长春:吉林大学,2014.
[2]李晗.基于机器视觉的高速车道标志线检测算法的研究[D].沈阳:东北大学,2006.
[2]李晗.基于机器视觉的高速车道标志线检测算法的研究[D].沈阳:东北大学,2006.