Multiscale Detection of Circles, Ellipses and Line Segments, Robust to Noise and Blur

This paper proposes a basic taxonomy of image contours. Our goal is to classify smooth curves into five categories, namely, circles, ellipses, line segments, arcs circles and ellipses. These geometrical structures have been chosen as they serve input many computer vision tasks. The proposed strategy applied on set initial disjoint contours, which are grouped together form the aforementioned structures. These, in turn, validated using an contrario approach that guarantees reduced number false detections. use multiscale permits detection at different resolution levels, makes method robust noise blur.