本文针对数字图像处理和计算机视觉中的圆检测问题，提出了基于半径/中心约束的模糊C-球壳聚类算法。首先利用共形几何代数理论以球壳替代传统的点作为聚类原型，以共形空间中模式与球壳的内积来表示模式与球壳的距离关系，并引入模糊隶属度函数来体现待分类模式和与之对应的球壳的隶属关系，给出了基于半径和中心约束的模糊C-球壳聚类算法的目标函数，最后给出了求解此类约束条件下目标函数最小化的统一方法。通过人工数据集和真实图像上的实验，均表明该算法能够有效地对球壳形数据进行模糊聚类，以及对数字图像进行圆检测。

A radius/center constraint Fuzzy C-Spherical Shell clustering (FCSS) algorithm is presented for the circle detection problem in digital image processing and computer vision. FCSS substitutes spherical shells for points as the prototype based on conformal geometric algebra, and the inner product of patterns and spherical shells in conformal space is equal to the distance from the patterns to spherical shells. Meanwhile, FCSS represents the membership between the patterns and its spherical shells by the introduction of fuzzy membership. Then the objective function for radius/center constraint FCSS clustering analysis is expressed, and a unified solution to minimize this constraint objective function is deduced. Experimental results of both artificial data and real image have proved that the clustering algorithm can cluster the spherical shell type data and detect circles in image effectively.