基于规则约束的耕地农作物类型遥感监测研究  

Remote sensing monitoring of crop types based on rule constraints

作　　者：张玉娟 孙恺 ZHANG Yujuan;SUN Kai(Shandong Provincial Institute of Land Surveying and Mapping,Ji'nan,Shandong 250102,China)

机构地区：[1]山东省国土测绘院,山东济南250102

出　　处：《测绘标准化》2025年第1期149-156,共8页Standardization of Surveying and Mapping

摘　　要：山东省气候多样、地形及耕地农作物的构成复杂,基于遥感技术进行耕地农作物分类提取面临着诸多挑战。针对该问题,本文提出了一种基于空间规则约束的耕地农作物分类提取方法,采用多源遥感影像逐级融合和顾及多元特征的分类提取手段,针对不同监测场景建立监测策略,有效克服了监测区域耕地地块破碎、同物异谱和异物同谱等监测难点。在试点区域进行耕地农作物分类提取试验,结果证明,该方法适用于山东省冬小麦、玉米、大蒜等多种常见农作物分类,单类地物分类提取精度最高可达95%以上。该方法已在相关项目中得到应用,有效促进了遥感技术在农业领域的推广。Shandong Province is characterized by a diverse climate,complex terrain,and a varied types of crops.The classification and extraction of crops based on remote sensing technology faces numerous challenges.This paper proposes a crop classification and extraction method based on spatial rule constraints to solve the mentioned problem.The method can be used to effectively overcome such difficulties existed in the monitoring area as fragmented farmland plots,same object but different spectra,and different object but the same spectrum by using multi-source remote sensing images to be fused step by step and considering multiple features for classification extraction,as well as establishing monitoring strategies for different monitoring scenarios.The classification and extraction experiments of crops have been carried out in the pilot area,and the results proved that the mentioned method is suitable for the classification of various common crops in Shandong Province,such as winter wheat,corn,and garlic,and the best extraction accuracy for individual class reaching over 95%.This method has been applied in relevant projects,it effectively promote the remote sensing technology in the agricultural fields.

关 键 词：农作物类型 分类 多源遥感影像 特征 规则约束 

分 类 号：P237.9[天文地球—摄影测量与遥感]