Arc-based evaluation and detection of ellipses

摘要

We propose the use of elliptic arcs as a reliable criterion for ellipse evaluation and reconstruction. A technique based on pixel connectivity is developed for detecting the endpoints and subtended angles of elliptic arcs. Model-based distance and angular connectivity are introduced for quick and meaningful estimation of elliptic arcs. An iterative algorithm for multiple-ellipse fitting based on arc detection is presented. Experimental results confirm the robustness and accuracy of the algorithm for the fitting of multiple overlapping ellipses.