Effective elliptic arc selection from connected edge points
Tomonari Masuzaki, Yasuyuki Sugaya · IPSJ Transactions on Computer Vision and Applications · 2016
Abstract Extracting edge points from an image and fitting ellipses to them is a fundamental technique for computer vision applications. However, since the extracted edge points sometimes contain non-elliptic arcs such as line segments, it is a very difficult to extract only elliptic arcs from them. In this paper, we propose a new method for extracting elliptic arcs from a spatially connected point sequence. We first fit an ellipse to an input point sequence and segment the sequence into partial arcs at the intersection points of the fitted ellipse. Next, we compute residuals of the fitted ellipse for all input points and select elliptic arcs among the segmented arcs by checking the curvatures of the residual graph. Then, we fit an ellipse to the selected arcs and repeat the above process until the selected arcs do not change. By using simulated data and real images, we compare the performance of our method with existing methods and show the efficiency of our proposed method.