基于双目立体视觉的倒车环境障碍物测量方法
发布时间：2019-07-18 18:11
【摘要】：倒车引发的交通事故是主要的城市交通安全隐患之一,本文针对倒车安全及障碍物检测中传统机器视觉倒车图像存在失真,以及距离感知精度低的问题,提出了一种基于双目立体视觉的倒车环境障碍物测量方法.首先根据双目标定理论获取摄像头内、外参数并分析摄像头畸变情况,使用标定参数对双目图像进行校正,运用极线约束使双目图像平行共面,通过双目图像视差和三角测量原理获取图像中各目标的实际坐标,利用固定单一物体计算测量距离与实际距离间的误差,进行双目立体视觉倒车障碍物检测测量有效性研究.实验结果表明:摄像机横坐标与实际测量距离基本吻合,图像校正结果比较理想,图像共面且极线对齐,能有效检测出后方障碍物,并有效提高倒车环境感知能力和倒车安全性能.
[Abstract]:Traffic accident caused by reverse car is one of the main hidden dangers of urban traffic safety. Aiming at the distortion of traditional machine vision reverse image in reverse safety and obstacle detection, as well as the low precision of distance perception, a method of obstacle measurement in reverse environment based on binocular stereo vision is proposed in this paper. Firstly, according to the binocular calibration theory, the internal and external parameters of the camera are obtained and the camera distortion is analyzed. The calibration parameters are used to correct the binocular image, the polar line constraint is used to make the binocular image parallel and coplanar, the actual coordinates of each target in the image are obtained through the parallax and triangulation principle of the binocular image, and the error between the measured distance and the actual distance is calculated by using a fixed single object. The effectiveness of binocular stereo vision reverse obstacle detection and measurement is studied. The experimental results show that the transverse coordinates of the camera are in good agreement with the actual measured distance, and the image correction results are ideal. The image coplanar and polar line alignment can effectively detect the rear obstacles, and effectively improve the reverse environmental perception ability and reverse safety performance.
【作者单位】： 西南交通大学交通运输与物流学院;西南交通大学机械工程学院;中国电子科技集团第三十二研究所;
【基金】：国家自然科学基金项目(51175443,61271341)~~
【分类号】：U463.6
本文编号：2516001
[Abstract]:Traffic accident caused by reverse car is one of the main hidden dangers of urban traffic safety. Aiming at the distortion of traditional machine vision reverse image in reverse safety and obstacle detection, as well as the low precision of distance perception, a method of obstacle measurement in reverse environment based on binocular stereo vision is proposed in this paper. Firstly, according to the binocular calibration theory, the internal and external parameters of the camera are obtained and the camera distortion is analyzed. The calibration parameters are used to correct the binocular image, the polar line constraint is used to make the binocular image parallel and coplanar, the actual coordinates of each target in the image are obtained through the parallax and triangulation principle of the binocular image, and the error between the measured distance and the actual distance is calculated by using a fixed single object. The effectiveness of binocular stereo vision reverse obstacle detection and measurement is studied. The experimental results show that the transverse coordinates of the camera are in good agreement with the actual measured distance, and the image correction results are ideal. The image coplanar and polar line alignment can effectively detect the rear obstacles, and effectively improve the reverse environmental perception ability and reverse safety performance.
【作者单位】： 西南交通大学交通运输与物流学院;西南交通大学机械工程学院;中国电子科技集团第三十二研究所;
【基金】：国家自然科学基金项目(51175443,61271341)~~
【分类号】：U463.6
【相似文献】
相关硕士学位论文 前1条
1 孙维毅;基于双目立体视觉的自主资源勘探车辆环境识别技术研究[D];吉林大学;2014年
，本文编号：2516001
本文链接：https://www.wllwen.com/kejilunwen/qiche/2516001.html
