基于形态学的海洋锋形态特征提取  

Extracting feature of ocean front based on mathematical morphology

作　　者：薛存金[1] 苏奋振[1] 周军其[2] 郭衍游[3] 仉天宇[1] 

机构地区：[1]中国科学院地理资源研究所资源环境信息国家重点实验室海洋GIS中心,北京100101 [2]武汉大学遥感信息工程学院,湖北武汉430079 [3]中国科学院海洋研究所,山东青岛266071

出　　处：《海洋科学》2008年第5期57-61,共5页Marine Sciences

基　　金：国家863计划资助项目(2002AA639400,2003AA637030,2003AA604040,2004AA639820)

摘　　要：根据形态学梯度理论:不同尺寸的结构元素适合不同空间尺度的特征提取,结构元素尺寸的过大或者过小都会达不到最佳的特征边缘提取效果,从结构元素的尺寸δ与海洋锋横断面宽度γ和海流流幅的空间尺度λ的理论关系图推导中,提出δ,γ和λ的最佳定量关系:δ=γ/2,且δ<=λ/2,旨在对海洋锋形态特征进行提取。最后,以黑潮和墨西哥湾流为实验数据,进行海洋锋形态特征的自动提取,实验结果表明该定量关系的正确性和有效性。There have been some difficulties to auto-extract ocean fronts from remote sensing fields. And in morphological theory, different sizes of structure element are suitable for feature extraction of different scales. The optimal effect by too large or too small size of structure element can not be obtained. This work discusses the relationship between size of structure element δ, the width of ocean fronts Y and the width of ocean current A. The theory analysis proves that the optimal quantitive relationship is δ=Y/2 and δ≤λ/2. On the base of the relationship, the Kuroshio and the Gulf Stream are to be taken as experiment data for edge extraction based on morphological gradient, and it was shown that the quantitive relationship is correct and effective.

关 键 词：海洋锋 结构元素尺寸 形态学梯度 特征提取 海洋地理信息系统 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]