一种改进的基于高斯分布拟合的提取标志点像素坐标方法  

An Improved Extracting Pixel Coordinate of Control Points Method Based on Gauss Distribution Fitting

作　　者：铁菊红[1] 彭辉[1] 

机构地区：[1]成都信息工程学院软件工程系,四川成都610225

出　　处：《计算机与现代化》2008年第4期58-60,共3页Computer and Modernization

摘　　要：提取标志点像素坐标是传统的摄像机标定中关键的步骤之一,其精度影响着摄像机标定的精度。对于由圆点构成的标定板,使用常用的方法提取标志点像素坐标在精度上存在不足。为此提出了改进的基于高斯分布拟合的方法,该方法识别出服从高斯分布的圆环域,再对圆环域内的点进行高斯曲面拟合,求得标志点像素坐标。经试验对比分析,此方法具有较高的精度,而且运算速度快。Extracting pixel coordinate of control poims is one of important methods in general camera calibration. The accuracy of control poim pixel coordinate extraction has the pronounced effect on camera＇ s calibration. The common techniques for extracting pixel coordinate of comrol points in the calibration board consisted of rounds are unsatisfactory under some conditions. A new improved method based on gaussian distribution fitting is put forward, which recognizes the annular mask that counters gaussian distribution and then calculates control poims pixel coordinates using gaussian surface fitting. Results of experiments show that this method is accurate and fast.

关 键 词：摄像机标定 标志点 像素坐标 高斯分布拟合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]