Using Graphics Hardware for Enhancing Edge and Circle Detection
Computer Architecture Department, University of Malaga, 29071 Malaga, Spain
Pattern Recognition and Image Analysis (2007), pp. 234-241
@article{ruiz2007using,
title={Using Graphics Hardware for Enhancing Edge and Circle Detection},
author={Ruiz, A. and Ujald{‘o}n, M. and Guil, N.},
journal={Pattern Recognition and Image Analysis},
pages={234–241},
year={2007},
publisher={Springer}
}
A broad family of problems in computer vision and image analysis require edge and circle detection. This paper explores the properties of the Hough transform for such tasks, improving them under a novel implementation on commodity graphics hardware. We demonstrate both a faster execution and a more reliable detection under different scenarios and a range of parameters selection. Overall, a consistent 3-6 acceleration factor and beyond is achieved for a 500 MHz GeForce 7950 GX2 graphics card versus a typical 3 GHz Pentium 4 dual-core CPU.
December 5, 2010 by hgpu
