红边波段在水稻生育期识别中的应用研究

Application of Red Edge Band in Identification of Rice Growth Stages

为了研究红边波段对水稻生育期识别的影响,利用无人机多光谱影像,基于随机森林(random forest,RF)和支持向量机(support vector machine,SVM)两种算法,对水稻的分蘖期、拔节孕穗期、抽穗扬花期、成熟期进行识别。结果表明,归一化差值红边指数(normalized difference red edge index,NDRE)、红边叶绿素指数(chlorophyll index-red edge,CIrededge)识别的总体精度明显高于归一化植被指数(normalized differential vegetation index,NDVI)、绿边叶绿素指数(chlorophyll index-green,CIgreen)、增强型植被指数(enhanced vegetation index,EVI)的总体精度。NDRE在拔节孕穗期的用户精度比NDVI低,而在分蘖期、抽穗扬花期、成熟期的识别精度比NDVI高。相比于CIgreen,CIrededge在抽穗扬花期识别精度更高;各波段组合识别水稻生育期的总体精度表现为RGB+RE>RGB+NIR+RE>RGB+NIR,在RGB+RE中加入近红外波段后,总体精度明显降低,相比于红边波段,近红外波段带有较多干扰信息。RGB+NIR和RGB+RE在分蘖期和成熟期识别效果相近,但RGB+RE在拔节孕穗期和抽穗扬花期的识别精度较高。

To study the influence of red edge band on the iden⁃tification of rice growth stages,we identify the tillering stage,jointing and booting stage,heading and flowering stage,and maturity stage of rice by random forests(RF)and support vector machine(SVM)with multispectral images obtained by unmanned aerial vehicle.The results show that the overall ac⁃curacy values of normalized difference red edge(NDRE)in⁃dex and chlorophyll index-green(CIrededge)are significantly higher than that of normalized differential vegetation index(NDVI),chlorophyll index-green(CIgreen)and enhanced vegetation index(EVI).The user accuracy of NDRE is lower than that of NDVI at jointing and booting stage,while its identification accuracy at tillering stage,heading and flowering stage,and maturity stage is higher than that of NDVI.Com⁃pared with CIgreen,CIrededge has higher identification accu⁃racy at heading and flowering stage.The overall accuracy values of band combinations for identifying rice growth stages show that RGB+RE>RGB+NIR+RE>RGB+NIR.Af⁃ter adding near-infrared band into RGB+RE,the overall ac⁃curacy is significantly reduced.Compared with the red edge band,the near-infrared band has more interference informa⁃tion.RGB+NIR and RGB+RE have similar identification effect at tillering stage and maturity stage,but RGB+RE has higher recognition accuracy at jointing and booting stage and heading and flowering stage.