数据驱动的二维点集自适应聚类算法

这里针对二维点集的聚类问题提出了一种新的自适应算法．该算法首先采用特征滤波器，提取样本集中具有线性特征的点，同时生成样本点之间的相互关系；然后根据相互关系，自适应地生成线状和球状分布的分类．实验结果表明，该算法不仅能够成功地对包含有多种分布的样本进行分类，而且具有较强的抗噪声能力．

An adaptive clustering algorithm for two-dimension objects

A new clustering algorithm is presented for two dimension objects. First, it uses a feature filter to obtain objects and their relations that have a linear feature. Then, it creates curves and clusters on basis of the relations. Experimental results show that the method presented not only succeeds in the clustering task but also is immune to noise.