Please use this identifier to cite or link to this item:
Title: Robust fitting of multiple curves
Authors: QIAO YU
Keywords: Linearly parameterizable model, weighted least squares approach, curve fitting, pixel connectivity, circle detection, ellipse detection
Issue Date: 31-May-2004
Citation: QIAO YU (2004-05-31). Robust fitting of multiple curves. ScholarBank@NUS Repository.
Abstract: The accurate detection of geometric shapes is an important problem in machine vision applications. This thesis describes a novel connectivity-based estimator for fitting multiple intersecting linearly parameterizable curves. The use of pixel connectivity effectively avoids false curve detection, improves the robustness against noise and significantly reduces the computational load. A robust single-model extractor is used for outlier detection. The underlying models can be extracted sequentially by repeatedly applying this extractor to the updated edge data set. Detected models are then evaluated with a model verifier to determine their validity. Three major distinguishing features of this estimator are: (i) robustness against outliers even when the number of outliers exceeds 50%, (ii) ability to extract an a priori unknown number of meaningful models; and (iii) effective fitting of multiple intersecting curves. We demonstrate the performance of the estimator by fitting intersecting or occluded circles and ellipses.
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 



Page view(s)

checked on Nov 25, 2018


checked on Nov 25, 2018

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.