密度聚类方法在船舶航迹图谱挖掘中的应用研究  

Research on the density clustering method in ship track pattern mining

作　　者：谢斌[1] 陈莉明[2] 

机构地区：[1]成都理工大学工程技术学院,四川乐山614007 [2]乐山师范学院,四川乐山614000

出　　处：《舰船科学技术》2016年第14期100-102,共3页Ship Science and Technology

基　　金：乐山市产学研合作资助项目(15GZD142)

摘　　要：研究密度聚类方法在船舶航迹图谱挖掘中的应用。针对航迹点密度不均匀的特性,给出一种改进的密度聚类方法,即结合分类的基于弥散度的聚类算法。该方法采用弥散度作为相对密度的度量标准,根据其动态分布来确定核心点与边缘点,并使用KNN核密度估计方法对边缘集合分类。实验结果表明,本文所使用的密度聚类方法能够对船舶航迹进行有效聚类。In this paper,the application of density clustering method in ship track pattern mining was studied. In order to solve the uneven density of track,an improved density clustering method was proposed,this method was a combination classification of the clustering algorithm. In this method,the dispersion degree was used as a measure of relative density,and the core point and edge point were determined according to its dynamic distribution,And KNN kernel density estimation method was used to classify the edge set. The experimental results shown that the density clustering method used in this paper can effectively cluster the ship tracks.

关 键 词：密度聚类 航迹图谱 弥散度 

分 类 号：U675.79[交通运输工程—船舶及航道工程]