摘要
选取场景中常见的门形结构为处理对象,提取门形结构中的三条直线,拟合直线方程,进而确定直线的交点及两条平行直线的消失点,根据消失点原理和垂直关系等先验知识确定摄像机坐标系和世界坐标系的映射。空间解析过程以直线方程为基本要素,参与解析运算的点特征均由直线方程确定,随机选取的点坐标仅用于方向一致性判定,减少了随机误差引入。仿真实验表明,本文的视觉定位方法对噪声有较高的鲁棒性,与融合线特征和随机点特征的方法相比降低了定位误差。
This paper selected common door-shaped structure in the scene as processing object.The method extracted the three straight lines from door-shaped structure image for fitting the straight lines equation,and determined the two intersections coordinate of the three straight lines and the vanishing point of two parallel straight lines.Then according to vanishing point principle and the vertical relationship,the map between the camera coordinate system and world coordinate system is calculated.Space analytic process was based on line features.The points,participated in analytic process,were calculated by line equation.Random points were only used to judge the direction consistency.All of them reduced the random errors.By simulation,it is shown that locating method in this paper has better robustness,and reduces position error compared with other method integrated line feature and random point feature.
出处
《光电子技术》
CAS
北大核心
2014年第2期78-83,共6页
Optoelectronic Technology
基金
国家自然科学基金资助项目(61374019)
南京工程学院青年基金重点项目(QKJA201305)
关键词
机器视觉
位姿估计
视觉定位
线特征
消失点
machine vision
pose estimation
visual localization
line feature
vanishing point