最小二乘改进算法及其在椭圆拟合中的应用  

Improved Least Square Algorithm and Application in Ellipse Fitting

作　　者：马向南[1] 李航[1] 刘丽丽[1] 刘志伟[1] 

机构地区：[1]河南科技大学机电工程学院,河南洛阳471003

出　　处：《河南科技大学学报（自然科学版）》2014年第3期18-21,5,共4页Journal of Henan University of Science And Technology:Natural Science

基　　金：河南省基础与前沿技术研究计划重大基金项目(082300413204)

摘　　要：提出一种像素级边缘检测椭圆拟合新算法,用该算法对最小二乘算法进行了改进。首先,将符合要求的准椭圆转化到归一化坐标系;然后利用最小二乘法进行亚像素级椭圆拟合;最后,采用二次曲线拟合点集求解出亚像素及椭圆几何中心。在给定的图形中,利用本文提出的改进像素级边缘检测算法可以明显提高拟合不确定度和拟合精度。A sub-pixel edge detection ellipse fitting algorithm to improve least squares algorithm was proposed.Firstly,the quasi-elliptic which met the requirements was translated into the normalized coordinate system.Secondly,the least squares method was used to fit ellipse with sub-pixel. Finally,the set of points was fitted by quadratic curve and the sub-pixel geometric center was solved. In a given graphic,the fitting uncertainty and precision are improved with the improved sub-pixel edge detection algorithm.

关 键 词：最小二乘法 边缘检测 椭圆拟合 亚像素 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]