光学技术, 2017, 43 (2): 138, 网络出版: 2017-04-10
基于邻域贡献权值细化的圆心亚像素定位算法
Sub-pixel location of circle based on neighbor contribution weight refinement
光学测量 邻域贡献权值 高斯拟合 非极大值抑制 边缘细化 亚像素 随机抽样一致性 optical measurement neighbor contribution weight Gaussian fitting non-maxima suppression edge thinning sub-pixel random sample consensus (RANSAC)
摘要
针对传统的圆心算法过程复杂、定位精度受初始边缘提取效果影响较大等问题, 提出了一种基于邻域贡献权值细化的圆心亚像素定位算法。首先引入邻域贡献权值系数, 改进传统非极大值抑制法, 细化边缘;然后在边缘点的梯度方向对灰度值进行高斯拟合, 确定亚像素边缘位置;最后针对边缘突变点提出了基于随机抽样一致的最小二乘法来拟合圆心。实验结果表明, 该算法具有较好的精度和稳定性, 圆心的提取精度可以达到0.1个像素。
Abstract
The traditional sub-pixel algorithm of edge detecting is complex, and the positioning accuracy changes because of the result of the initial edge detecting, a method is proposed. The neighbor contribution weight coefficient is led to improve the non-maxima suppression method and the edge is refined, and then the exact edge is obtained by Gaussian fitting of the gray value in the direction of the gradient, finally the discontinuity points are rejected and the center is fitted by the random sample consensus method. The experiment result shows, the proposed algorithm is of better accuracy and robustness, and the extracting accuracy of the center is to 0.1 pixels.
游江, 唐力伟, 邓士杰, 苏续军. 基于邻域贡献权值细化的圆心亚像素定位算法[J]. 光学技术, 2017, 43(2): 138. YOU Jiang, TANG Liwei, DENG Shijie, SU Xujun. Sub-pixel location of circle based on neighbor contribution weight refinement[J]. Optical Technique, 2017, 43(2): 138.