一种基于高斯混合模型的海上浮标轨迹聚类算法  

A Clustering Algorithm for Sea Buoy Trajectory Based on Gaussian Mixture Model

作　　者：荆晓刚[1] 葛丽阁 孙伟[2] 

机构地区：[1]上海港引航站,上海200082 [2]上海海事大学信息工程学院,上海201306

出　　处：《现代计算机》2017年第24期3-5,8,共4页Modern Computer

摘　　要：海上环境不同于陆地,其不受道路、轨道的限制和受表面风流场多因素影响,其目标的运动轨迹更显杂乱,给海上目标的轨迹分析带来挑战。提出一种基于高斯混合模型的海上浮标轨迹的聚类算法。该算法将高斯混合模型应用于漂移浮标的复杂不规则轨迹的聚类,能够有效消除轨迹中异常点的影响。仿真实验表明针对浮标漂移轨迹GMM算法较K-means算法更优,鲁棒性更好。该研究成果可应用于海上搜救、航路规划等领域。The sea environment is different from the land trajectory has a certain regularity, a variety of factors are not controlled, resulting in mari- time trajectory analysis is more difficult. So, presents an algorithm for sea buoy trajectory clustering based on Gaussian Mixture Model. The algorithm can be applied to the unrestricted and complex trajectory of the sea buoy, and the irregular trajectory. And it can effectively elimi- nate the influence of abnormal points in the trajectory. The experimental results show that the Gaussian mixture model clustering algorithm has higher reliability than K-means. The research results can be applied to such fields as maritime search and rescue, route planning and

关 键 词：聚类 高斯混合模型 浮标 漂移轨迹 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]