作 者：翟陆续[1,2] 李毅[1,2] 肖朝[1,2]
ZHAI Lu-xu LI Yi, XIAO Chao （1. College of Computer Science, Sichuan University, Chengdu 610065, China; National Key Laboratory of Fundamental Science on Synthetic Vision, Sichuan University, Chengdu 610065, China）
机构地区：[1]四川大学计算机学院,四川成都610065 [2]四川大学视觉合成图形图像技术国防重点学科实验室,四川成都610065
出 处：《计算机工程与设计》 CSCD 2014年第35卷第9期 3190-3195页,共6页
Computer Engineering and Design
基 金：基金项目：国家自然科学基金项目（60903118、60832011）;国家科技支撑计划基金项目（2012BAH62F03）
摘 要：针对传统几何校正的不足,提出一种改进的几何校正的方法,只需少许人工干预就能很高精度的完成几何校正.介绍一种相机自动定位和拍照的方法,避免人工调整相机位置的繁琐工作,实现相机控制的自动化;提出一种基于递归的去除图片公共噪声区域算法,更精确地识别出特征点;通过临近的特征点,使用平行四边形插值算法与最小二乘法曲线拟合算法对被遮挡住的特征点坐标进行估计.实验结果表明,相机控制方法提高了几何校正的自动化程度,去噪算法提高了识别特征点的精度,平行四边形插值算法可以很好地对被遮挡的特征点坐标进行估计。
An auto-geometric calibration method was proposed which avoided the defects of the traditional geometric calibration and needed only a little manual intervention to achieve high precision. Firstly, for realizing the automated camera control, a method of auto-camera position and auto-camera photographing was described which avoided tedious manual work to adjust the camera position. Then a recursive algorithm for picture noise removal was introduced, so as to identify the target point more ac- curately. Finally, to estimate the position of the obscured target point, the parallelogram interpolation algorithm, the least square method and points which were adjacent to the obscured point were used. Experimental results show that the auto-camera control method improves the automation degree of the geometric calibration, the noise removal algorithm improves the recognition accuracy of the target point, and the parallelogram interpolation algorithm can estimate the coordinates of the obscured point accurately.
关键词：几何校正 相机自动定位 去噪算法 特征点识别 特征点位置估计
geometric calibration ; auto-camera position ; noise removal algorithm ; target recognition ; target point position estimation ;
改进的多投影仪显示墙几何校正技术.pdf
(955.19 KB, 下载次数: 0)