基于图像的大豆作物覆盖度快速提取方法研究  

Research on Fast Extraction Method of Soybean Crop Coverage Based on Images

作　　者：马旭 杨继文 邓蜀江 季毅民 MA Xu;YANG Jiwen;DENG Shujiang;JI Yimin(The Second Institute of Geographic Information Cartography of the MNR,Harbin 150081,China;Harbin Map Publishing House Co.,Ltd.,Harbin 150081,China)

机构地区：[1]自然资源部第二地理信息制图院,黑龙江哈尔滨150081 [2]哈尔滨地图出版社有限公司,黑龙江哈尔滨150081

出　　处：《测绘与空间地理信息》2024年第S01期15-18,共4页Geomatics & Spatial Information Technology

摘　　要：本研究选取了东北典型大豆主产区的样地为试验区,利用RGB阈值法、植被指数(NDI、VEG、CIVE、EXG、EXR、EXGR、COM)对开花期、成熟期获取的图像进行植被覆盖度提取,评估各提取方法的适用性。研究表明:本研究的9种提取方法精准度均达到89%以上,对于叶片呈深绿色的大豆生长期,9种提取方法均具有较高的精准度(94%以上);而对于叶片呈黄绿色的成熟期,EXG、EXRG、CIVE、VEG和COM指数具有较好的适用性。In this study,the typical soybean production area in Northeast China was selected as the experimental area.The RGB threshold method and vegetation index(NDI,VEG,CIVE,EXG,EXR,EXGR and COM)were used to extract the vegetation coverage from the images acquired at flowering and maturity stages,and the applicability of each extraction method was evaluated.It was shown that the accuracy of the nine extraction methods in this study was higher than 89%.For the growth period of dark green soybean leaves,the nine extraction methods had high accuracy(more than 94%),while EXG,EXRG,CIVE,VEG and COM possessed better versatility compared with the other methods for the mature stage of yellow green leaves.

关 键 词：植被覆盖度 大豆作物 数字图像 

分 类 号：P237[天文地球—摄影测量与遥感]