Concentric circle detection based on normalized distance variance and the straight line Hough transform
Xing Chen, Ling Lu, Sheng Yang · 2014
Without any special qualifications, this paper proposed a novel method for concentric circles detection by using the geometrical characteristics of circle and the straight line Hough transform. A feature point neighborhood method and a normalized distance variance method are proposed to segment continuous curves and circles respectively for removing interference of noncircular objects which simplified the computation; Thirdly, circle centers and radius are detected by the straight line Hough transform. Because only two 2-dimensional voting are required for the circles detection, the proposed method avoided the problem of large storage which exists in the classical Hough transform and the detection efficiency is also enhanced. Finally, the concentric circles are identified and located according to the relation between the positions of circles. Experimental results demonstrate that, compared with the classical Hough transform, the proposed method improved execution efficiency greatly while maintaining high detection precision, it also offers good robustness in some complex images.