基于TM遥感影像溢油识别方法的比较研究  

Comparative Analysis of Oil Spill Recognition Based on TM Remote Sensing Image

作　　者：陈韩 谢涛 徐辉 孟雷 陈伟[5] CHEN Han;XIE Tao;XU Hui;MENG Lei;CHEN Wei

机构地区：[1]南京信息工程大学海洋科学学院,江苏南京210044 [2]南京信息工程大学遥感与测绘工程学院,江苏南京210044 [3]国家气象中心,北京100081 [4]北京市5111信箱,北京100094 [5]武汉理工大学信息工程学院,湖北武汉430070

出　　处：《武汉理工大学学报（信息与管理工程版）》2019年第3期305-311,共7页Journal of Wuhan University of Technology：Information & Management Engineering

基　　金：国家自然科学基金项目(41776181);国家重点研发计划基金项目(2016YFC1401007;2018YFC1506404);江苏省研究生科研创新计划基金项目(KYCX18_1012)

摘　　要：为了及时掌握溢油信息,实现对溢油区域的高精度监测,采用平行管道法、最大似然法和支持向量机法3种监督分类方法识别两幅TM影像的溢油区域,分析各种分类法在溢油识别领域的差异和其适用范围,同时利用绿波段和反射红外波段的有效组合对溢油区域的范围加以确定。结果表明:支持向量机分类法在两景TM遥感影像溢油识别上的总体精度和Kappa系数均为最高,过分割误差与欠分割误差较低,且时间复杂度较为合理,能够对大面积溢油实现较好的监测。In order to timely grasp the oil spill information and achieve high-precision monitoring of the oil spill area, the supervised classification method (parallel hexahedron method, maximum likelihood method and support vector machine method) is used to identify the oil spill areas of the two TM images, and analyzes various classifications. The difference between the method in the field of oil spill identification and its scope of application, while using the effective combination of the green band and the reflected infrared band to determine the range of the oil spill area. The results show that the overall accuracy and Kappa coefficient of the SVM classification method are the highest in the two-view TM image oil spill recognition. Over-segmentation error and under-segmentation error are low, and the time complexity is reasonable, which can achieve better monitoring of large-area oil spill.

关 键 词：污染监测 监督分类 支持向量机 海面溢油 最大似然法 

分 类 号：P76[天文地球—海洋科学]