Accelerating Circle Detection Based on Generalized Projection Method with GPUs

Shape recognition, an important portion of image processing, has been widely applied in many research and application areas. Traditional Hough transform, which is a modern shape recognition technique, incurs unbearable execution time and memory storage cost during the extraction of high-dimensional shapes. This paper presents an approach for circle extraction based on generalized image projection and parallelizes it on several brands of GPUs. Experiments shows that GPUs can achieve up to almost 500 times speedup with this method over single threaded CPUs, while traditional Hough transform can only achieve about 100 times speedup. Massive parallelization capability and broad memory bandwidth make GPUs feasible for practical use. Combined with some efficient image scaling algorithms, real-time processing of circle extraction can be achieved.

[1]  Antonio Ruiz,et al.  Using Graphics Hardware for Enhancing Edge and Circle Detection , 2007, IbPRIA.

[2]  J. Radon.  On the determination of functions from their integral values along certain manifolds , 1986, IEEE Transactions on Medical Imaging.

[3]  Ignacy Duleba,et al.  Circular Object Detection Using a Modified Hough Transform , 2008, Int. J. Appl. Math. Comput. Sci..

[4]  Zhang Rui-lin.  Circle detection using Hough transform , 2008 .

[5]  Dana H. Ballard,et al.  Generalizing the Hough transform to detect arbitrary shapes , 1981, Pattern Recognit..

[6]  Nader Bagherzadeh,et al.  A New Approach for Circle Detection on Multiprocessors , 1994, J. Parallel Distributed Comput..

[7]  H. Mizumoto,et al.  Binocular robot vision system with shape recognition , 2007, 2007 International Conference on Control, Automation and Systems.

[8]  Takeo Kanade,et al.  Detection of hematopoietic stem cells in microscopy images using a bank of ring filters , 2010, 2010 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.

[9]  Richard O. Duda,et al.  Use of the Hough transformation to detect lines and curves in pictures , 1972, CACM.