Very fast ellipse detection using GPU-based RHT

An approach to very rapid computation of one class of randomized Hough transform (RHT) using parallel processing capabilities of a programmable graphics processing unit is described. The method is able to detect ellipses in real-time, even in large images. It uses fragment processing in recovering ellipse shape parameters. Its effectiveness is evaluated through experiments on synthetic and real images.

[1]  Steve Mann,et al.  OpenVIDIA: parallel GPU computer vision , 2005, ACM Multimedia.

[2]  Marcus A. Magnor,et al.  A graphics hardware implementation of the generalized Hough transform for fast object recognition, scale, and 3D pose detection , 2003, 12th International Conference on Image Analysis and Processing, 2003.Proceedings..

[3]  Ling-Hwei Chen,et al.  A high-speed algorithm for elliptical object detection , 1996, IEEE Trans. Image Process..

[4]  Robert A. McLaughlin,et al.  Randomized Hough Transform: Improved ellipse detection with comparison , 1998, Pattern Recognit. Lett..

[5]  Antonio Ruiz,et al.  On the computation of the Circle Hough Transform by a GPU rasterizer , 2008, Pattern Recognit. Lett..

[6]  Zhi-Qiang Liu,et al.  A robust, real-time ellipse detector , 2005, Pattern Recognit..

[7]  Erkki Oja,et al.  Randomized Hough transform (RHT) , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.

[8]  Timothy S. Newman,et al.  Shape-based mechanisms for content-based retrieval of aurora images , 2006, SPIE Optics East.