Robust metrological characterization of circular profiles

Author:

Greco L,Caja J,Maresca P

Abstract

Abstract Outliers are likely to corrupt data acquired by optical profiles and ignoring them can badly affect the output of the analysis. The occurrence of very few outliers can completely invalidate the fitting process and uncertainty evaluations. Outliers are unexpected values that often mask each other being not detectable by any pre-screening of the data. Therefore, there is the need to adopt a proper robust approach to fit circular profiles data and identify data inadequacies. Here, it is suggested to resort to impartial trimming to fit the centre and the radius of a circle. The optimization problem is based on a subset of the data after discarding that fraction of points with the largest distance from the fitted centre. Trimmed observation are likely to be outliers. The methodology is illustrated through both synthetic and real data.

Publisher

IOP Publishing

Subject

General Medicine

Reference11 articles.

1. Metrological characterization of interior circular features using digital optical machines. Calculation models and application scope;Caja;Precision Engineering,2014

2. Comparison of methods for outlier identification in surface characterization;Yang;Measurement,2018

3. Random sample consensus: a paradigm for model fitting with application to image analysis and automated cartography;Fischler;Commun. Assoc. Comp. Mach.,1981

4. Circle fitting by linear and non-linear least squares;Coope;Journal of Optimization Theory and Applications,1993

5. Approximate Maximum Likelihood Estimation of Circle Parameters;Chan;J Optim Theory Appl,2005

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3