复杂道路环境中前方车辆检测方法

发布时间：2018-04-02 18:14

本文选题：支持向量机　切入点：毫米波雷达　出处：《科学技术与工程》2017年33期

【摘要】：为提高前方车辆检测在不同道路环境中的鲁棒性和实时性,提出一种基于支持向量机的多传感器融合前方车辆检测方法。系统工作前利用多传感器数据融合建立雷达坐标与图像坐标的转化关系,以毫米波雷达在各种复杂道路环境中前方障碍物的检测数据为基础,利用支持向量机(SVM)训练分类器构建车辆与非车辆识别系统,最终根据车辆宽高比的统计规律,建立前方车辆识别窗口。道路试验结果表明该方法前方车辆识别准确率为89.2%,单帧图像的处理速度为31 ms。对于不同道路环境中的前方车辆检测表现出了良好的稳定性和准确性,总体性能取得较为显著的提高。
[Abstract]:In order to improve the robustness and real-time performance of forward vehicle detection in different road environments, a multi-sensor fusion forward vehicle detection method based on support vector machine (SVM) is proposed.Before the system works, multi-sensor data fusion is used to establish the transformation relationship between radar coordinates and image coordinates, which is based on the detection data of obstacles in front of millimeter-wave radar in various complex road environments.The support vector machine (SVM) training classifier is used to construct the vehicle and non-vehicle recognition system. Finally, according to the statistical rule of the aspect ratio of the vehicle, the forward vehicle recognition window is established.The road test results show that the accuracy of vehicle recognition in front of the method is 89. 2 and the processing speed of single frame image is 31 Ms.It shows good stability and accuracy for the forward vehicle detection in different road environments, and the overall performance can be improved significantly.
【作者单位】：山东理工大学;
【基金】：国家重点研发计划(2016YFD0701101) 山东省自然科学基金(ZR2016EL19)资助
【分类号】：TP391.41;U463.6;U495

【相似文献】