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

作　　者：魏本赞 田言亮[2] 卢辉雄[1,3] 牛海威 董双发[1] 张建永[1,3] 

机构地区：[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]东华理工大学地球科学学院,江西南昌330013

出　　处：《安徽农业科学》2016年第6期313-315,318,共4页Journal of Anhui Agricultural Sciences

基　　金：中国地质调查局资助项目(1212011220941)

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。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 

分 类 号：S126[农业科学—农业基础科学]