面向对象的海域使用专题信息遥感提取关键技术研究  

Key technologies study of object-oriented thematic information extraction for sea area usage by remote sensing

作　　者：谢伟军 韩飞[2] 张东[2] 汪闽[2] 崔丹丹 吕林 朱瑞 

机构地区：[1]江苏省海域使用动态监管中心,江苏南京210003 [2]南京师范大学地理科学学院,江苏南京210023

出　　处：《海洋环境科学》2014年第2期274-279,共6页Marine Environmental Science

基　　金：国家海洋局海域管理技术重点实验室开放基金项目(201207);海洋公益性项目(20100511)

摘　　要：利用遥感技术进行海域使用专题信息提取是开展区域用海项目动态监测的有效手段。本研究以SPOT-5高空间分辨率遥感影像为数据源,探讨了面向对象的海域使用专题信息遥感提取方法,从基于多特征的遥感图像多尺度分割、海域使用特征基元的多特征表达以及面向对象的海域使用支持向量机SVM(Support Vector Machine)分类三个方面,深入研究了面向对象的海域使用专题信息遥感提取关键技术与方法,结合所提出的海域使用两级分类体系,形成了利用高分遥感技术实现区域用海遥感监测的技术流程。在江苏如东沿海进行的实证分类研究表明,该技术流程能够取得理想的分类结果,可为开发业务化的海域使用遥感动态监测平台系统提供技术支撑。Extraction of sea area usage thematic information by remote sensing method is an effective way for dynamic monitoring of regional sea use project to express its variational trends. Taking SPOT-5 high spatial resolution remote sensing image as basic data source,objective of this paper is to thoroughly study key technologies for information extraction of sea area usage via remote sensing image processing. Here three kinds of techniques and methods are involved which include： multiscale segmentation technique with the support of multi-feature remote sensing image; multiple feature expression method for sea-use feature units and object-oriented SVM（ Support Vector Machine） method for image classification. They are then sequentially composed the technical processes for information extraction of regional sea areas. These processes are applied on coastal area near Rudong country,Jiangsu province and get ideal experimental classification results. So it could become the supporting techniques for the construction of ocean dynamic monitoring platform. This platform could be applied for business usage in daily sea area usage monitoring and management works.

关 键 词：面向对象 海域使用 多尺度分割 特征基元 SVM分类 

分 类 号：X87[环境科学与工程—环境工程]