摘要
分析和比较了三种常见的基于矩的亚像素边缘定位方法,得出三种方法获取的边缘角度以及OFMM法与ZOM法获取的亚像素边缘相同的结论。提出了一种针对椭圆目标的亚像素边缘定位方法。在ROI提取和SOBEL算子初步检测边缘的基础上,利用初步获取的椭圆目标的几何信息,将参与计算的模板个数减少至一个,从而大大减少了运算时间,提高了算法的运算速度。同时实验结果表明,与其他基于矩的亚像素边缘定位方法相比,新方法不仅有更高的运算速度,而且精度和抗噪性能都有很大地提高。
Analysis and comparisons for three common methods for sub-pixel edge location based moment are made.The results of edge-angle computed by the three moment-based methods are all the same.Moreover,the results of sub-pixel edge location computed by OFMM method and ZOM method are also the same.A novel method for sub-pixel edge of ellipse is proposed.Based on ROI extraction and coarsely edge-location by SOBEL operator,the geometric information of the target is used to reduce the number of moment-template to be only one,which can largely save the time.From the results both in simulation and practical experiments,the proposed method not only has greater computing speed,but also has a great advantage over other methods in precision and anti-noises.
出处
《光学技术》
CAS
CSCD
北大核心
2010年第1期66-74,共9页
Optical Technique
基金
国家自然科学基金资助项目(60775025)
江苏省自然科学基金资助项目(BK2007116)
关键词
光学测量
矩
边缘定位
亚像素
图像处理
optical measurement
moment
edge location
sub-pixel
image processing
作者简介
张虎（1985-），男，硕士研究生，从事光学测量与图像处理研究。E-mail： h13077395959@hotmail. com
