基于机载多光谱遥感数据的溢油信息提取方法  

Method for oil spill information extraction from airborne multispectral imagery

作　　者：刘丙新[1] 李颖[1] 高超[1] 

机构地区：[1]大连海事大学环境信息研究所,大连116026

出　　处：《国土资源遥感》2014年第1期42-46,共5页Remote Sensing for Land & Resources

基　　金：国家自然科学基金项目"船舶溢油目标光谱重建和识别研究"(编号:41071260);国家海洋公益项目"海洋溢油污染风险评估及应急响应关键技术集成和示范应用"(编号:2012418012-02);中国石油天然气股份有限公司科学研究与技术开发项目"遥感物化探油气直接检测和评价技术研究"(编号:2011A-0209-01)共同资助

摘　　要：为研究应用紫外-可见光-近红外-热红外遥感数据识别海上溢油的方法,利用机载多光谱溢油监测设备获取了舟山海域多光谱遥感数据,并利用野外地物光谱仪获取了水体和油膜的反射光谱。通过提取并分析机载多光谱遥感图像上多目标的影像特征和光谱响应特征,利用决策树分类法对海上油膜相对厚度的分布进行分类和制图。研究表明,基于机载多光谱遥感数据的溢油信息提取方法总体分类精度达93.7%,能够准确区分薄油膜和厚油膜,可有效提取海洋溢油污染信息,完全能够满足海洋溢油污染遥感监测需求。In order to study the method for identifying the oil film using the multispectral data, the authors obtained the imagery of Zhoushan waters by using the marine aircraft - borne multi - spectral oil spill detector, and also acquired the reflectance spectra of the water and oil film by means of FieldSpec spectroradiometer. The image characteristics and spectral response characteristics were analyzed so as to detect the distribution of the relative thickness using the decision tree classifier. The overall accuracy is 93.7%, which shows that the classification method based on the spectral characteristics could effectively recognize the thick thin sheen oil film and extract the information of marine oil pollution, thus sufficiently satisfying the requirements for monitoring marine oil spill.

关 键 词：机载多光谱数据 光谱特征 溢油 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]