多模式组合的大区域地块级农作物分类方法研究  

Research on Large Area and Plot Level Crop Classification Methods Based on Multi-model Combination

作　　者：张力仁 冯然 吴长俊 马学民 ZHANG Liren;FENG Ran;WU Changjun;MA Xuemin(The Second Institute of Surveying and Mapping Engineering of Heilongjiang Province,Harbin 150025,China)

机构地区：[1]黑龙江第二测绘工程院,黑龙江哈尔滨150025

出　　处：《测绘与空间地理信息》2024年第S01期49-51,55,共4页Geomatics & Spatial Information Technology

摘　　要：农作物精细分类信息是农作物监测的基础数据,随着遥感技术在农作物分类理论、技术、方法方面取得长足发展,基于多源多时序遥感影像的农作物智能识别得到越来越广泛的应用。对于大范围的地块级农作物种植结构调查而言,目前受限于训练样本不足、缺乏完备的农业调查统计体系等因素,遥感技术尚无法全面支撑农作物调查的工程化应用。针对目前智慧农业、精准农业管理等实际工作需求,本文提出了以“农业专题数据”为基底定量,以高分辨率遥感影像智能解译为基础,结合实地调查与人工编辑的一种“多模”组合下大范围的地块级农作物种植结构调查方法,并以齐齐哈尔市为例进行应用验证。利用混淆矩阵对农作物分类结果进行精度验证,总体精度和Kappa系数分别为94.66%和0.91,分类结果精度较高。通过验证,此种方法可准确获取地块级农作物分类空间数据,可以为精准农业、智慧农业管理提供准确数据支撑。Crop fine classification information is the basic data of crop monitoring.With the rapid development of remote sensing technology in the theory,technique and method of crop classification,intelligent crop recognition based on multi-source and multi-time series remote sensing images has been applied more and more widely.For large-scale and plot-level crop planting structure investigation,remote sensing technology is still unable to fully support the engineering application of crop investigation due to the lack of training samples,lack of complete agricultural investigation statistical system and other factors.In view of the current practical requirements of smart agriculture and precision agriculture management,this paper proposed a“multi-model”combination investigation method of large-scale and plot-level crop planting structure based on intelligent interpretation of high-resolution remote sensing images,field investigation and manual editing by using"agricultural thematic data"as basic quantification and took Qiqihar city as an example for application verification.Confusion matrix was used to verify the accuracy of crop classification results.The overall accuracy and Kappa coefficient were 94.66%and 0.91,respectively,indicating high accuracy of classification results.Through verification,this method can accurately obtain the spatial data of crop classification at plot-level,which can provide accurate data support for precision agriculture and smart agriculture management.

关 键 词：多模式组合 大区域 地块级 农作物分类 

分 类 号：P237[天文地球—摄影测量与遥感]