基于骨架提取的船舶航迹聚类技术研究

Research on ship track clustering based on skeleton extraction

利用用于模型重构的骨架提取技术,对船舶历史数据进行轨迹聚类分析,为研究船舶行为模式奠定基础,进而对区域航行物监管提供新方法。针对目前轨迹聚类算法消耗计算资源大,无法快速处理航迹线的问题,将航迹线转换成图像进行梯度压缩和抽取聚类。依靠热力距离场构建航迹线的热曲面,后利用拉普拉斯算子对网格化的热曲面进行迭代收缩,得到剖面骨架线作为聚类效果图。通过获取我国东南沿海的船舶自动识别系统(Automatic Identification System,AIS)数据进行仿真实验,并可视化呈现。结果表明,将骨架提取技术应用到航迹聚类中,在达到预期聚类效果的情况下,可以避免处理大量的复杂雷达定位点数据,从而较大缩短聚类计算时间。

Using the skeleton extraction technology for model reconstruction,the trajectory clustering analysis of ship history data is carried out,which lays a foundation for the study of ship behavior pattern and provides a new method for the supervision of regional navigation objects.Aiming at the problem that the current track clustering algorithm consumes a lot of computational resources and cannot process track lines quickly,the trajectory is converted into an image for gradient compression and extraction clustering.The thermal surface of track line is constructed based on the thermal distance field,and then the meshed thermal surface is iteratively contracted by Laplace operator,and the skeleton line is obtained as the clustering effect.Simulation experiment is carried out by obtaining the data of Automatic Identification System(AIS)of ships in the southeast coast of China,and visual presentation is conducted.The results show that when the skeleton extraction technology is applied to the track clustering,the expected clustering effect can be achieved,and a large number of complex radar registration data can be avoided to reduce the clustering calculation time.