一种改进的光帧差移动目标跟踪与检测算法研究  

A new algorithm of moving target tracking and detection based on optical flow techniques with frame difference methods

作　　者：刘远仲 张海波[1,2] 杨嘉 唐天国[2] 谭鹤毅[2] LIU Yuanzhong;ZHANG Haibo;YANG Jia;TANG Tianguo;TAN Heyi(Nanchong Electronic Information Industry Institute of Technology,Nanchong 637000,China;Nanchong Professional Technic College,Nanchong 637000,China)

机构地区：[1]南充电子信息产业技术研究院,四川南充637000 [2]南充职业技术学院,四川南充637000

出　　处：《电子设计工程》2021年第4期139-144,149,共7页Electronic Design Engineering

基　　金：南充市科技局研发资金项目(19YFZJ0022);四川省教育厅自然科学研究课题(16ZB0466)。

摘　　要：在传统移动目标的跟踪与检测算法基础上,提出了一种改进的智能视频跟踪与检测算法,目的是要提高普通视频中移动目标跟踪技术的实时性、有效性。为了降低传统视频中移动目标跟踪的编程难度,减小计算量,在前期采用图像帧差法来捕捉移动目标的区域,在后期运用最优估计点匹配技术与光流均匀采样技术构建光流场实现图像的检测与跟踪。实验仿真表明,该方法能对视频流中的移动目标进行实时有效的跟踪与检测。This paper introduces an improved intelligent video tracking and detection algorithm based on the traditional moving target tracking and detection algorithm,the purpose is to improve the ordinary video moving target tracking technology in real-time,effectiveness.In order to reduce the traditional moving target tracking in video programming difficulty,reduce the amount of calculation,based on the previous image frame differential method is used to capture the area of moving targets,in the later use of optimal estimation point matching technology and optical flow uniform sampling technique to build an image of a light flow field to realize the detection and tracking of,through the experimental simulation show that this method can effectively the moving targets in video streams of real-time effective tracking and detection.

关 键 词：光流场 图像帧差法 移动目标检测与跟踪 梯度算法 

分 类 号：TN911.73[电子电信—通信与信息系统]