本文为美国克莱姆森大学（作者：Vinay Gidla）的硕士论文，共41页。

车辆跟踪分类是计算机视觉的一个重要应用。跟踪技术主要有两种：基于外观的跟踪和基于特征的跟踪。基于外观的技术通常需要跟踪目标的模型或模板，然而，该模型需要对目标运动过程中可能出现的大量变形具有鲁棒性。

本研究探讨了使用基于视图的模板集合，而不是单一模板，来精确追踪交通运动中拖车的轮廓。跟踪从创建模板序列开始，通过手动处理包含所需类型牵引车的交通段视图。处理过程包括手动将拖车与每帧中的背景场景分割开来。模板序列旨在捕捉车辆在车道上行驶时所有可能姿态下的外观，然后，该集合用于检测跟踪类似拖车在车道上运动的过程。为了在跟踪过程中更好地对准轮廓，引入了诸如梯度幅度信息等显著特征。

Vehicle tracking and classification is animportant application of computer vision. There exist mainly two types oftracking techniques: Appearance-based tracking and Feature-based tracking.Appearance-based techniques generally require a model or template of the targetto be tracked. However the model needs to be robust to the multitude ofdeformations possible in the course of the target’s movement. This research exploresthe use of a collection of view based templates, instead of a single template,to accurately trace the contours of the tractor-trailers in moving traffic. Thetracking begins with creation of a template sequence by manually processing asegment of traffic which contains an exemplar of the desired class oftractortrailers. The processing involves manually segmenting thetractor-trailer from the rest of the scene in each frame. The template sequenceis intended to capture the appearance of the vehicle in all possible poses asit moves through the lane. This collection is then used to detect and tracksimilar tractor-trailers during their movement through the lane. Salientfeatures, such as gradient magnitude information, are incorporated for betteralignment of the contour during tracking.

1 引言

1.1 道路交通分析中的计算机视觉

1.2 以前工作的回顾

1.3 跟踪方法

1.4 本文概要

2 使用模板序列进行基于视图的模板跟踪

2.1 算法概述

2.2 模板序列创建

2.3 背景减除

2.4 模板选择

2.5 模板匹配小结

2.6 算法小结

3 实验结果

3.1 跟踪器框架的结果

3.2 虚假结果

3.3 基于梯度幅度匹配的影响

4 结论

更多精彩文章请关注公众号：