摘要
随着内河船舶通航密度加大,面向水上船舶智能化安全监管,提出了一种多源异构传感器融合的船舶目标检测及动态跟踪方法,重点分析了图像目标检测和多源数据融合算法。针对图像目标检测,提出了基于边缘检测的三帧差分法与基于混合高斯背景减除法相结合的船舶视频目标检测算法;针对多源异构数据融合,优化了一种正态性隶属度函数计算模糊矩阵的方法,实现了在动态视频修正下的激光点云数据与AIS数据的船舶轨迹特征融合,并通过空间矩阵变换实现投影到同一坐标系,最后利用卡尔曼滤波算法实现了目标的动态跟踪。通过该方法研发了一套基于多传感器融合的船舶态势主动式智能感知系统,经系统分析,该方法比传统人工检测及单一监测手段都具有更好的环境适用性和检测精度。
With the increase of navigable density of inland river ships,a ship target detection and dynamic tracking method based on multi-source and heterogeneous sensor fusion is proposed for the intelligent safety supervision of water-based ships,and the algorithm of image target detection and multi-source data fusion is emphatic analyzed.Aiming at image target detection,a ship video target detection algorithm which combines three frame difference method based on edge detection and Gaussian background subtraction method was proposed.For multi-source heterogeneous data fusion,this paper optimized a normality membership function calculation method of fuzzy matrix,realized under the dynamic video correction laser point cloud data with the ship trajectory characteristics of AIS data fusion,and through space matrix projection to the same coordinate system,and finally by using kalman filter algorithm to achieve the goal of dynamic tracking.Through this method,a set of ship situation active intelligent sensing system based on multi-sensor fusion is developed.Experimental verification shows that this method has better environmental applicability and detection accuracy than traditional manual detection and single monitoring means.
作者
马瑞鑫
李子龙
陈静
MA Rui-xin;LI Zi-long;CHEN Jing(Tianjin Research Institute for Water Transport Engineering,M.O.T.,Tianjin 300456,China;Key Laboratory of Marine Simulation and Control,Dalian Maritime University,Dalian 116026,China)
出处
《水道港口》
2021年第3期392-398,共7页
Journal of Waterway and Harbor
基金
天津市交通运输科技计划项目(22118061)
中央级公益性科研院所基本科研业务费项目(TKS20210301)。
关键词
船舶目标检测
船舶目标跟踪
数据融合
计算机视觉
激光点云
vessel target detection
vessel target tracking
data fusion
computer vision
laser point cloud
