Combined AGADESN with DBSCAN Algorithm for Cluster Target Motion Intention Recognition

Algorithm 1

DBSCAN algorithm flow.

Input: Uncertain data set D, Clustering radius Eps, Minimal number of neighboring points MinPts

Output: A set of clusters, types of all the points in D

Main function of the algorithm:

  DBSCAN(D, Eps, MinPts)

    ClusterNum=0

  for each target point P in D

    ifP is visited

    continue to next point

    end if

  mark P as visited

    Eps-Neighborhood=regionQuery(P, Eps)

    if sizeof(Eps-Neighborhood)<MinPts

     mark P as noise

    else

    ClusterNum=next cluster

    expandCluster(P, Eps-Neighborhood, ClusterNum, Eps, MinPts)

    end if

   end for

  END

  regionQuery(P, Eps)

     returnEps-Neighborhood(P)={Q∈S|D(P,Q)≤Eps}

    expandCluster(P, Eps-Neighborhood, ClusterNum, Eps, MinPts)

    add P to cluster ClusterNum

     for each point in Eps-Neighborhood

      if is not visited

       mark as visited

        =regionQuery(, Eps)

       if sizeof()>=MinPts

        Eps-Neighborhood=Eps-Neighborhood joined with

       end if

      end if

      if is not yet member of any cluster

       add to cluster ClusterNum

      end if

end for