UNA APROXIMACIÓN AL PROBLEMA DE LA DETECCIÓN DE ELIPSES EN PRESENCIA DE RUIDO Y VALORES ATÍPICOS

Abstract

We propose a new approach for ellipses detection in scattered data. The method is constructed on the assumption that it is possible to estimate the tangents to the ellipse in given points. From this information the coefficients of the quadratic form of the ellipses can be recovered. On this base two univariate cost functions are defined. The minimization of the proposed functions produces robust results in the presence of outliers, something difficult to obtain with least squares methods. With the aim of reducing the computational cost, one of the proposed functions approximates the intended ellipse with a poligonal produced by an efficient subdivision scheme