基于雷视融合的高速公路交通检测方法

Highway Traffic Detection Method Based on Ray-Vision Fusion

针对单传感器检测的不足和复杂的道路环境,本文设计了视觉传感器与毫米波雷达的车辆目标检测融合算法。通过传感器的参数标定与坐标转换构建了空间融合模型,首先获取了传感器的内参及实际安装的位置信息,然后将毫米波雷达的坐标变换到图像的像素坐标,实现了传感器的时间融合。实验结果表明,与单一传感器的检测结果相比,视觉传感器与毫米波雷达的融合算法能够显著改善视频杂波、雷达的多点分布以及检测目标的分割等问题,目标检测的准确率高于单一传感器,特别是在复杂场景中,融合算法检测的稳定性更高。

Aiming at the insufficiency of single-sensor detection and the complex road environment,this paper designs a fusion algorithm for vehicle target detection based on visual sensors and millimeter-wave radar.The spatial fusion model is constructed through the parameter calibration and coordinate transformation of the sensors,and the internal reference of the sensors and the position information of the actual installation are obtained firstly,and then the millimeter-wave radar’s coordinates are transformed to the pixel coordinates of the image to realize the temporal fusion of the sensors.The experimental results show that the fusion algorithm of visual sensors and millimeter-wave radar can significantly improve the problems of video clutter,multipoint distribution of radar,and segmentation of detected targets compared with the detection results of a single sensor,and the accuracy of target detection is higher than that of a single sensor,especially in complex scenes,the fusion algorithm has a higher stability of detection.