Irregular Ellipse Detection Method by Confined Space Sweep

A method for the detection of elliptical figures with irregular contours in images is proposed. This method overcomes the problem of heteroclite contours that causes the dispersion of the parameters during the accumulation of votes in the Hough parametric domain, effect that is usually reduced by increasing the discretization step of the parameters, although with the consequence of reducing the precision of the results. With the use of an inter ellipses region of interest (IEROI), an ellipse pixel search area is defined, and in turn, the geometric characteristics of the irregular object are delimited. This IEROI can be adapted to detect, both, irregular and uniform contours. Similar to the Hough transform, the parameters of the detected ellipses are stored in a 5-dimensional parametric space, but the traditional voting method is replaced by the use of a value that corresponds to the Confinement Grade of the object contour within the IEROI. Finally, the functionality of the proposed method is validated for different cases, using real and synthetic images with uniform and irregular elliptical contours.