摘要
现有基于点云与图像融合的行人检测要求高算力的处理平台,应用于低算力低功耗的嵌入式平台时,无法满足行人检测的准确率和实时性。基此提出一种融合点云与图像的道路行人检测方法,该方法采用DBSCAN算法对点云进行聚类,然后,运用概率数据关联算法将行人点云与图像的行人检测结果进行决策级融合,最后,在嵌入式计算平台上进行软硬件集成与测试验证。实验结果表明,相比于其他目标检测算法,设计的融合点云与图像的道路行人检测方法,不仅提高了道路行人方位的检测精度,而且检测用时降低了46.6%以上。
The existing road pedestrian detection methods based on point cloud and visual fusion require a processing platform with high computing power.When applied to the embedded platform with low computing power and low power consumption,they cannot produceaccurate and real-time pedestrian detection.To solve this problem,this paperproposesa pedestrian detection method fusing point cloud and image.The DBSCAN algorithm is used to cluster the point cloud.Then,the pedestrian detection results of pedestrian point cloud and image are fused at the decision level based on the probabilistic data association algorithm.Finally,the software and hardware integration and test verification are carried out on the embedded computing platform.The experimental results show that compared with other target detection algorithms,the detection accuracy of pedestrians on the road is improved and the detection time is reduced by more than 46.6%by the pedestrian detection method fusing point cloud and image designed in this paper.
作者
王长海
陈倩
唐欣
叶进
WANG Chang-hai;CHEN Qian;TANG Xin;YE Jin(Guangxi Research Institute of Comprehensive Transportation Big Data,Nanning 530001,China;School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China;Guangxi Communications Design Group Co.,Ltd.,Nanning 530001,China;School of Computer,Electronics and Information,Guangxi University,Nanning 530004,China;Guangxi Key Laboratory of Multimedia Communications and Network Technology,Nanning 530004,China)
出处
《广西大学学报（自然科学版）》
CAS
北大核心
2021年第6期1592-1601,共10页
Journal of Guangxi University（Natural Science Edition）
基金
国家自然科学基金资助项目(61762030)
广西科技计划项目(AA18242021,ZY19183005,AB20238033,AB21196032)
桂林市科技计划项目(20190214-3)
广西高校中青年教师基础能力提升项目(2021KY1654)。
关键词
点云
图像检测
行人检测
概率数据关联
决策级融合
智能驾驶
point cloud
image detection
pedestrian detection
probabilistic data association
decision-level fusion
intelligent driving
