一种基于多传感器信息融合的盲区检测与目标跟踪方法

A Blind Area Detection and Target Tracking Method Based on Multi-Sensor Information Fusion

随着经济和社会的发展,大型商用车保有量逐年增加,因其车辆构造产生视野盲区导致的交通事故也随之增加。传统单一类型传感器检测精度低,已无法满足如今复杂交通场景的应用。基于此,文章提出了一种基于多传感器信息融合的盲区检测与目标跟踪方法,采用多种传感器感知盲区环境,基于YOLOv3算法与DeepSORT算法,改进提出自适应的快速DBSCAN聚类算法,利用CIOU匹配替换传统IOU匹配,提升目标聚类精度与算法匹配性能,实现更高精度的盲区目标检测与跟踪。

With the development of the economy and society,the number of large commercial vehicles has been increasing year by year,and the traffic accidents caused by blind areas in the field of view due to their vehicle structure have also increased.The traditional single type of sensor detection accuracy is low and can no longer meet the application of today’s complex traffic scenarios.Based on this problem,the article proposes a blind zone detection and target tracking method based on multi-sensor information fusion,using multiple sensors to sense the blind environment,improving the adaptive fast DBSCAN clustering algorithm based on YOLOv3 algorithm and DeepSORT algorithm,and replacing the traditional IOU matching with CIOU matching to improve the target clustering accuracy and algorithm matching performance,and achieving higher accuracy of blind spot target detection and tracking.