面向对象影像分类与提取法在农作物遥感解译中的应用
Application of Object-oriented Classification and Extraction Method in Remote Sensing Interpretation of Crop
魏本赞 1田言亮 2卢辉雄 3牛海威 1董双发 1张建永3
作者信息
- 1. 核工业航测遥感中心,河北石家庄050002
- 2. 中国地质科学院水文地质环境地质研究所,河北石家庄050061
- 3. 核工业航测遥感中心,河北石家庄050002;东华理工大学地球科学学院,江西南昌330013
- 折叠
摘要
根据Landsat8遥感数据,利用eCognition软件,以黄淮海平原饶阳县某地区为试验区,通过对分割参数的定量试验,确定了地物对象的最佳分割尺度;在此基础上分析植被、非植被的典型特征,采用面向对象方法研究植被的信息提取,并对提取结果进行了精度评价.试验结果表明:面向对象影像分类方法能极大地提高分类精度,并实现了植被信息的准确提取,为我国主要粮食基地农作物遥感调查与监测提供了技术支撑.
Abstract
According to the Landsat8 remote sensing data,quantitative test on splitting parameter was carried out by eCognition software,with a given area in Raoyang County of Huang-Huai-Hai Plain as the research region.The optimal splitting scale of earth object was determined.Based on these,typical characteristics of vegetation and non-vegetation were analyzed.The information extraction of vegetation was researched by object-oriented method.Extraction results were precisely evaluated.Research results showed that object-oriented image classification method greatly enhanced the classification accuracy,realized the accurate extraction of vegetation information,and provided technical support for the crop remote sensing investigation and monitoring of main grain base.
关键词
Landsat8/面向对象/影像分割/信息提取/精度评价Key words
Landsat8/Object-oriented/Image segmentation/Information extraction/Precision evaluation引用本文复制引用
基金项目
中国地质调查局资助项目(1212011220941)
出版年
2016
国家科技期刊平台
NSTL
维普
万方数据