Affiliation:
1. Wuhan University of Technology
2. Hubei University of Technology
Abstract
As the traditional Hough transform has such defects as large storage space and long computing time in ellipse detection, an improved randomized ellipses detection method based on least squares was presented, which utilizes the least square approach to fit the ellipse and combines both of the advantages of the random Hough transform and the least square. By setting appropriate distance threshold of the candidate ellipse and the threshold of edge points, the method of ellipse detection decreases the number of random sampling and the invalid calculation of cumulation in the process of Hough transform. The results show that the method doesn’t require large storage space, has good ability to overcome the noise and realizes the fast detection for the single ellipse and defective ellipse.
Publisher
Trans Tech Publications, Ltd.
Reference5 articles.
1. Bennett, Burridge and Saito. A methed to detect and characterize ellipses using the Hough transform. IEEETPAMI: IEEE Transaction on Pattern Analysis and Machine Intelllgence, v21: 652-657, (1999).
2. HO Chun-ta, CHEN Ling-hwei. A fast ellipse /circle detector using geometric symmetry. Pattern Recognition, 28(1): 117-124, (1995).
3. H.T. Sheu, H.Y. Chen and W.C. Hu. Consistent symmetrical axis method for robust detection of ellipses. IEE Proceedings-Vision Image and signal Processing, 144(6): 332-338, (1997).
4. QU Wen-tai. Chord midpoint Hough transform based ellipse detection method. Journal of Zhe jiang University (Engineering Science), 39 (8): 1132-1135, (2005).
5. Yang Zhong-gen, Cao Fang. Subspace projection and QR decomposition applied to ellipse fitting. ICSP 2006, v3, (2006).