摘要
为实现对地学信息单元的自动识别和分类,对地学信息单元的几何形状特征的提取及量测方法进行了研究,利用数学形态学方法实现对遥感图像的半自动提取地学信息单元的图形边缘.分别提取图形的面积、周长、面积周长比、形状参数、分形分维数等5个直观易测且具有判别意义的特征参数.利用这5个特征采用判别分析方法实现对地学信息单元的识别分类,并取得了较理想的结果和准确率.
In order to realize the automated recognition and classification of the Geo-info-unit, firstly we do some research on the abstraction and measures of the Geo-info-unit and get the outline of the Geo-info-unit based on the remote sensing image by the mathematical morphology; then calculate the parameters of the Geo-info-unit, such as the area,perimeter,the ratio of area to perimeter,shape index,fractal dimension, which can be computed simply and used to fulfill the classification of the Geo-info-unit.
出处
《武汉理工大学学报（交通科学与工程版）》
2008年第6期1021-1024,共4页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
国家863计划项目资助(批准号:2001AA135140)
东华大学青年教师科研基金项目资助(批准号:113-10-0044026)
关键词
数学形态学
地学信息单元
遥感图像
特征提取
识别
mathematical morphology
geo-info-unit
remote sensing image
abstraction
classification
