利用图谱特征解析和反演作物叶绿素密度  

Inversion and evaluation of crop chlorophyll density based on analyzing image and spectrum

作　　者：张东彦[1,2] 刘良云[3] 黄文江[3] Cobum Craig 梁栋[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039 [2]北京农业信息技术研究中心,北京100097 [3]中国科学院数字地球重点实验室,北京100094 [4]Department of Geography, University of Lethbridge, Alberta T1K3M4, Canada

出　　处：《红外与激光工程》2013年第7期1871-1881,共11页Infrared and Laser Engineering

基　　金：中国科学院数字地球重点实验室资助(2012LDE003);安徽省高等学校省级自然科学研究项目(KJ2013A026);安徽省自然科学基金青年基金(1308085QC58);国家自然科学基金(41071228、61172127);中国博士后科学基金(2013T60189);安徽大学博士科研启动经费

摘　　要：地面成像光谱仪可对作物个体及群体信息进行"图谱"同步解析,因此在农业定量化研究中具有巨大的应用潜力。利用可见-近红外成像光谱仪采集不同生育期玉米和大豆的冠层"图谱"数据,在逐步提取影像中光照土壤、阴影土壤、光照植被、阴影植被四种组分光谱的基础上,通过选取的敏感波段构建光谱植被指数和叶绿素密度进行波段自相关分析,探讨各个分量对作物叶绿素密度反演的影响。研究发现:当植被与土壤混合存在时,对叶绿素密度敏感的波段基本在红光与近红外波段;当植被光谱提纯后(剔除土壤光谱),对叶绿素密度敏感的波段范围增大,表现在蓝、绿波段;当阴影叶片光谱剔除后,对叶绿素密度敏感的波段表现为可见光波段增加,近红外波段减少,红边波段决定系数最高。上述变化特征在不同作物中有相同的趋势,为探索地面成像光谱仪"图谱"协同反演作物生化参数进行了有意义的探索。Field imaging spectrometer can be used to analyze growth information of individual and groupcrop relying on its data advantage with combination image and spectra as one, so it has great applicationpotential in agricultural quantitative research. In this research, hyperspectral images of com and soybeanin different growth period were collect using visible and near-infrared imaging spectrometer （VNIS）, andspectra of four components as illuminated soil, shadow soil, illuminated vegetation and shadow vegetation were gradually extracted, then spectral vegetation index was constructed based on different sensitive bands. On the basis, through analyzing bands correlation between chlorophyll density and spectral vegetation index, those influences for different components on chlorophyll density inversion of crop were explored. Some results can be found that when spectral information came from mixed canopy including vegetation and soil, sensitive bands for chlorophyll density were red light and near-infrared light. When soil spectra was removed, sensitive bands enlarged and showed in blue and green light region, and when spectra of shadow leaves were removed, sensitive bands indicated that visible light bands increased and near-infrared light bands decreased, there was the highest determination coefficient in red light region. Those change characteristics had same trend in different crops, this paper has important meaning for exploring inversion of biochemistry parameters on crop using data with combination image and spectra as one.

关 键 词：图谱解析 光谱指数 叶绿素密度 大豆 玉米 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] S127[自动化与计算机技术—控制科学与工程]