【作者】 王辉；

【导师】 徐亚明；

【作者基本信息】 武汉大学 ， 大地测量学与测量工程， 2005， 硕士

【摘要】 图像的特征获取是三维重构、运动跟踪、图像分割等工作的前提条件,在摄影测量及计算机视觉中有着重要的地位。本文重点研究了三视张量在图像点特征及线特征获取中的应用。 三视张量是三视几何的数学表达,它对三幅图像上的同名点、同名线都存在严格的约束关系。它能将目标同名点限制于一个点上,将目标同名线限制于一条线上。利用三视张量能够消除两视情况下特征点、特征线匹配算法所不能发现的错误匹配。 本文研究了数字图像的去噪,特征点、线的提取以及特征点、线的匹配。其中,特征点、线的匹配问题引入了三视张量约束条件。最后,设计了一套图像特征获取系统。实验表明,该系统能较好的获取图像特征点、特征线。更多还原

【Abstract】 The feature tracking in images is the preconditions of 3D reconstruction, motion tracking and image segmentation, it is important in digital photogrammetry and computer vision. This paper mainly researches the application that one uses the trifocal tensor to get feature points and feature lines.The trifocal tensor is the mathematical expression of trifocal geometry, it has strict constraints to correspondence points and lines in three images. It can restrict the object correspondence points to one point and the object correspondence lines to one line. The trifocal tensor can eliminate the wrong matches while the epipolar geometry algorithms of feature point and line match can not find them.This paper researches denoise, tracking of feature points and feature lines and matching of feature points and feature lines in digital images. Thereinto, the problem of matching feature points and feature lines introduce the trifocal tensor as the constraint condition. In the end, the paper designes a system of getting image features. Experiments demonstrate that the system can preferably get feature points and feature lines.更多还原