DBScan 算法

DBScan 是一种基于密度的聚类算法，主要算法流程如下图：

DBSCAN(D, eps, MinPts)
   C = 0                                          //类别标示
   for each unvisited point P in dataset D        //遍历
      mark P as visited                           //已经访问
      NeighborPts = regionQuery(P, eps)           //计算这个点的邻域    
      if sizeof(NeighborPts) < MinPts             //不能作为核心点
         mark P as NOISE                          //标记为噪音数据
      else                                        //作为核心点，根据该点创建一个类别
         C = next cluster
         expandCluster(P, NeighborPts, C, eps, MinPts)    //根据该核心店扩展类别
          
expandCluster(P, NeighborPts, C, eps, MinPts)
   add P to cluster C                                     //扩展类别，核心店先加入
   for each point P' in NeighborPts                       //然后针对核心店邻域内的点，如果该点没有被访问，
      if P' is not visited
         mark P' as visited                               //进行访问
         NeighborPts' = regionQuery(P', eps)              //如果该点为核心点，则扩充该类别
         if sizeof(NeighborPts') >= MinPts
            NeighborPts = NeighborPts joined with NeighborPts'
      if P' is not yet member of any cluster              //如果邻域内点不是核心点，并且无类别，比如噪音数据，则加入此类别
         add P' to cluster C
          
regionQuery(P, eps)                                       //计算邻域
   return all points within P's eps-neighborhood

结合百度百科的伪代码：