摘要
基于空间点集的连通性构造的等价关系,提出一种针对大规模数据集的快速分组算法。该算法的本质仍是基于等价关系的集合划分算法,但在关系矩阵传递闭包的构造算法中,提出一种快速的实现方法。与传统等价关系的构造方法相比,新算法需要较少的存储空间,而且算法复杂度由O(n3)降为O(km2)(k n,m≈n)。在医学断层图像表面重建的应用中证实,新算法具有重要的应用价值。
Based on the equivalent relationship constructed from the connectivity, a fast grouping algorithm is presented for mass data set. Compared with the traditional method of equivalent relationship construction, the proposed algorithm requires less memory and has the computational complexity of O(km^2) rather than O(n^3), where k is far less than n and m approximately equals to n. The new algorithm has been applied to the advance post-processing of the 3D surface rendering for medical images with satisfied results in objects' cutting, preserving and deletion.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2004年第1期106-109,共4页
Systems Engineering and Electronics
基金
北京大学视觉与听觉信息处理国家重点实验室基金资助课题
关键词
等价关系
数据处理
图像分析
equivalent relationship
date processing
image analysis
