Asymptotically efficient estimators for the fittings of coupled circles and ellipses

被引：0

作者：

Ma, Z. ^{[1
]}

Ho, K.C. ^{[1
]}

机构：

[1] Department of Electrical and Computer Engineering, University of Missouri, Columbia,MO,65211, United States

来源：

Digital Signal Processing: A Review Journal | 2014年 / 25卷

关键词：

Image processing - Parameter estimation - Geometry - Linear transformations - Benchmarking - Gaussian noise (electronic);

D O I：

暂无

中图分类号：

学科分类号：

摘要：

Fitting a pair of coupled geometric objects toa number of coordinate points is a challenging and important problem in many applications including coordinate metrology, petroleum engineering and image processing. This paper derives two asymptotically efficient estimators, one for concentric circles fitting and the other for concentric ellipses fitting, based on the weighted equation error formulation and non-linear parameter transformation. The Kanatani-Cramer-Rao (KCR) lower bounds for the parameter estimates of the concentric circles and concentric ellipses under zero-mean Gaussian noise are provided to serve as the performance benchmark. Small-noise analysis shows that the proposed estimators reach the KCR lower bound performance asymptotically. The accuracy of the proposed estimators is corroborated by experiments with synthetic data and realistic images. © 2013 Elsevier Inc. All rights reserved.

引用

页码：28 / 40