摘要
本文基于均值漂移方法设计了一种新的多目标被动连续跟踪算法。该算法分为两部分:先是对采集到的多目标图像进行预处理,包括灰度化、消噪和背景抑制三项工作,从根本上提高目标图像本身的质量;然后融合特征提取和目标检测步骤,利用均值漂移算法实现对多目标物体的检测与跟踪。仿真实验结果表明,与几种典型跟踪算法相比,应用基于均值漂移的算法后,多目标跟踪误差更小,其跟踪结果的平均偏差最大值仅为1.61 m,且跟踪响应耗时最大值仅为900 ms,证明其能够实现对运动目标的准确、可靠跟踪。
In the paper,a new multi-target passive continuous tracking algorithm is designed based on mean shift method.The algorithm is divided into two parts.First,the collected multi-target image is preprocessed,including grayscale,noise elimination and background suppression,so as to fundamentally improve the quality of the target image itself.Then the steps of feature extraction and target detection are fused and the mean shift algorithm is used to detect and track multi-target objects.The simulation results show that compared with several typical tracking algorithms,the multi-target tracking error is smaller after applying the algorithm based on mean shift,the maximum average deviation of the tracking result is only 1.61 m,and the maximum tracking response time is only 900 ms,which proves that the algorithm can achieve accurate and reliable tracking of moving targets.
作者
张博
龙慧
Zhang Bo;Long Hui(College of Information Science and Engineering,Changsha Normal University,Changsha 410100,China)
出处
《单片机与嵌入式系统应用》
2021年第10期27-31,35,共6页
Microcontrollers & Embedded Systems
基金
教育部产学合作协同育人项目(201901014024)
湖南省教育厅科学研究项目重点项目(20A036)
校级教学改革项目(JG2020025)。
关键词
均值漂移算法
多目标连续跟踪
消噪
背景抑制
mean shift algorithm
multi-target continuous tracking
de-noising
background suppression
