Robust detection of several circles or ellipses with heuristics

In this article we present a new approach to extract points that belongs to several ellipses or circles presented on a same image and with the presence of outliers. Each geometric form is extracted by means of a robust fitting, that is a nonlinear optimization problem, solved with two different heuristics: differential evolution and RANSAC. Once the geometric form is fitted, its points are extracted by calculating their statistics. Several tests with synthetic and real images are performed to show its effectiveness.