关于目标追踪算法的相关研究综述
首发时间:2020-11-20
摘要:目标追踪是在给定视频序列初始帧的目标大小和位置的情况下,预测后续帧中该目标的大小与位置,形成目标物体的一段运动轨迹。随着科技的发展,人们对于算法的性能要求越来越高,更快的速度和更精确的精度是目前的研究重难点,本文按照追踪算法的出现时间对各时期的代表性算法进行了梳理和介绍,本文首先介绍了追踪算法的分类组成和研究难点,进而针对研究难点从经典追踪算法、基于相关滤波的追踪算法和基于深度学习的追踪算法3个方面对代表性算法进行了归纳和分析,其中在基于相关滤波算法的基础上,分别从经典滤波算法、尺度变换、边界效应以及颜色直方图和置信度等4个常见问题介绍了代表性的相关滤波算法,重点梳理了每一种算法的原理和适用场景。最后简单的介绍了现有的公开数据集并且对于目标追踪算法的未来发展趋势进行展望,为使用目标追踪算法的相关研究人员提供一份参考。
For information in English, please click here
A survey of related research on target tracking algorithm
Abstract:Target tracking is to predict the size and position of the target in subsequent frames given the target size and position of the initial frame of the video sequence and form a motA survey of target tracking algorithmsion trajectory of the target object.Along with the development of science and technology, people are increasingly demanding for the performance of the algorithm, the faster and more accurate precision is the current study difficult point, time line according to the tracking algorithm in this paper generalizes the typical algorithm of each period and introduction, this paper first introduces the tracking algorithm of classification and research difficulties, then the research from the classic tracking algorithm, based on relevant filter tracking algorithm and tracking algorithm based on depth study three aspects has carried on the induction and analysis of typical algorithm, which on the basis of the algorithm based on correlation filtering,The typical correlation filtering algorithm is introduced from four common problems including classical filtering algorithm, scale transformation, boundary effect, color histogram and confidence degree, and the principle and applicable scenes of each algorithm are emphatically summarized.In the end, the existing open data set is briefly introduced and the future development trend of target tracking algorithm is prospected, which provides a reference for researchers using target tracking algorithm.
Keywords: target tracking Computer vision Algorithm overview
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
关于目标追踪算法的相关研究综述
评论
全部评论0/1000