Ellipse detection with elemental subsets

Abstract

We propose a simple method for fitting ellipses to data sets. The method first computes the fitting cost of small samples, called elemental subsets. We then prove that the global fitting cost can be easily derived from the fitting cost of the samples. Since fitting costs are computed from small samples, the technique can be incorporated in many ellipse detection and recognition algorithms, and in particular, in algorithms that make use of incremental fitting. Some of the theoretical results are formulated in the more general setting of implicit curve fitting. © 2009 Springer Berlin Heidelberg.