基于遥感图像的船舶交通流量预测方法  

Ship traffic flow prediction method based on remote sensing images

作　　者：曾晓晴 曹慧[2] ZENG Xiao-qing;CAO Hui(Jiangsu Shipping College Institute,Nantong 226010,China;Nantong University,Nantong 226019,China)

机构地区：[1]江苏航运职业技术学院,江苏南通226010 [2]南通大学,江苏南通226019

出　　处：《舰船科学技术》2023年第17期162-165,共4页Ship Science and Technology

基　　金：江苏航运职业技术学院科研课题(HYKY/2020B03)。

摘　　要：为了保证船舶的安全运行,资源的合理利用,避免海上交通安全事故的发生,提出基于遥感图像的船舶交通流量预测方法。结合主成分分析法、多层元胞更新方法等建立船舶遥感图像的显著模型,获取船舶遥感图像的显著图;通过恒虚警(CFAR)检测方法在船舶遥感图像的显著图上进行遍历,完成船舶的目标检测;依据船舶目标检测结果,统计船舶数量,计算海上船舶的拥堵率,并建立支持向量机(SVM)模型,将船舶的拥堵率作为SVM模型的输入,与输出数据的结果进行映射,实现船舶交通流量的预测。实验结果表明:该方法可准确地预测出船舶交通流量;该方法不受海杂波等干扰的影响,可准确地检测出船舶目标;在云雾和海杂波的环境干扰下,该方法预测船舶交通流量结果的均等系数趋近于0,预测能力较强。In order to ensure the safe operation of ships,rational use of resources and avoid the occurrence of maritime traffic accidents,a ship traffic flow prediction method based on remote sensing images is proposed.The salient model of ship remote sensing image is established by combining principal component analysis and multi-layer cell updating method,and the salient map of ship remote sensing image is obtained.Constant false alarm(CFAR)detection method is used to traverse the significant map of the ship remote sensing image to complete the target detection of the ship.According to the ship target detection results,the number of ships is counted,the congestion rate of ships at sea is calculated,and the support vector machine(SVM)model is established.The congestion rate of ships is used as the input of the SVM model,and the results of the output data are mapped to realize the prediction of ship traffic flow.The experimental results show that this method can accurately predict the ship traffic flow.The method is not affected by sea clutter and can accurately detect the ship target.Under the environment interference of cloud,fog and sea clutter,the uniformity coefficient of the forecast results of ship traffic flow is close to 0,and the prediction ability is strong.

关 键 词：遥感图像 船舶显著图 目标检测 船舶拥堵率 船舶交通流量 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]