亚像素定位算法精度评价方法的研究  

Study on the Method of the Accuracy Evaluation of Sub-Pixel Location Operators

作　　者：刘国栋[1] 刘炳国[1] 陈凤东[1] 胡涛[1] 

机构地区：[1]哈尔滨工业大学光电信息测控技术与装备研究所,黑龙江哈尔滨150001

出　　处：《光学学报》2009年第12期3446-3451,共6页Acta Optica Sinica

基　　金：中国博士后基金(20060400820)资助课题

摘　　要：通过模拟实际光学成像过程以及CCD采样过程,生成了标准图像,实现对亚像素定位算法的客观评价,并提出了一种新的基于灰度面积插值的亚像素定位方法。经实验验证,该方法对直线边缘的平均定位精度达到0.11 pixel,对圆边缘点的平均定位精度可达0.24 pixel,优于梯度质心法和灰度矩法。The paper proposes a method to build standard images by simulating the process of optical imaging and CCD sampling, which provides an objective standard for the evaluation of sub-pixel location operators. Also a novel sub-pixel location operator based on gray area interpolation is proposed, experiments prove that the average accuracy of the arithmetic is about 0.11 pixel for linear edge, and 0.24 pixel for circle edge, which is better than the operators based on gradient centroid and gray moment.

关 键 词：亚像素定位 标准图像 计算机仿真 视觉测量 灰度矩 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]