基于Hough变换的直线跟踪方法  

A New Method of Lines Tracking Based Hough Transform

作　　者：陈震[1] 高满屯[2] 杨声云[1] 

机构地区：[1]南昌航空工业学院人工智能与图像处理研究中心,江西南昌330034 [2]西北工业大学机电工程学院,陕西西安710072

出　　处：《计算机应用》2003年第10期30-32,共3页journal of Computer Applications

基　　金：国家自然科学基金(60275037);航空基础科学基金(99F53065);江西省测试技术与控制工程研究中心开放基金(2001-15)

摘　　要：文中提出一种基于Hough变换的直线跟踪新方法,该方法利用Hough变换中的点线对偶性原理,将在图像空间中对直线的跟踪转化为在参数空间中对点的跟踪,把参数空间中的点作为特征点,在参数空间中的特征窗口内跟踪特征点,由最小二乘法得到匹配的特征点。文中还给出了该直线跟踪方法的几何解释,即在参数空间中的矩形特征窗口内搜索匹配点,实际上是在图像空间中以原点为中心的扇环范围内搜索匹配的垂足。合成图像序列和真实图像序列实验结果表明,该方法取得了较好的跟踪结果。This paper presents a new tracking lines method based on Hough transform. Rather than tracking lines in image space directly,the presented method makes use of duality theory in Hough transform in order to transform linetracking in image space into pointtracking in Hough space. In other words,the point parameter and in Hough space can determine a line uniquely in image space. The transformed point in Hough space is defined as feature point,so the feature point can be tracked by using KanadeLucasTomasi feature point track algorithm. This paper also presents the geometrically explanation of linetracking method,namely searching the matching point in the feature window that is actually to find the matched perpendicularity in part of a sector in image space. Experimental results using real image sequences show that the new method provides a good estimation of the lines tracking.

关 键 词：HOUGH变换 直线跟踪 点线对偶性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]