基于多角度图谱解析的小麦白粉病诊断方法研究

项目名称： 基于多角度图谱解析的小麦白粉病诊断方法研究

项目编号： No.41301471

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 天文学、地球科学

项目作者： 张东彦

作者单位： 安徽大学

项目金额： 25万元

中文摘要： 目前，基于光学遥感的作物病害诊断方法，大都是在垂直观测方式下获取作物冠层的光谱数据，进而构建病害遥感监测模型实现的。常规方法在小麦白粉病感染初期或潜伏期监测效果较差或根本无效。基于此，项目针对小麦白粉病自下而上的发病特征，采用多角度观测装置搭载地面成像光谱系统FISS，重点研究：1）通过图谱协同解析小麦白粉病早期/潜伏期的影像数据，确定表征病害特征的最优观测角度或角度组合；筛选出对病害敏感的波段或波段区间，实现对早中期病害特征的图谱识别；2）在此基础上，利用线性、非线性分析方法，建立融合多角度信息的病害预测模型并优化；3）通过低空遥感平台获取多角度影像，对多角度病害模型进行扩展性检验与验证。从而完善小麦白粉病的遥感监测方法，同时为多角度观测服务于大尺度的病害监测提供方法支持。

中文关键词： 高光谱成像；小麦白粉病；多角度；特征提取；模型

英文摘要： Currently, diagnosis models of crop disease based on optical remote sensing are mostly built using crop canopy spectral data under vertical observation. Conventional methods have not been appropriate for monitoring wheat powdery mildew when it is at early stage or incubation stage. Based on this, the project will utilize multi-angle observation device with Field Imaging Spectrometer System (FISS) to research infection characteristics of wheat powdery mildew from bottom to top. Through collaboratively analyzing spectra and images of wheat powdery mildew collecting from early stage or incubation stage, to determine the optimal angle or angle combination which can response disease features; to pick up sensitive bands or band range for disease features, to achieve recognition of spectra and images for disease characteristics at early and medium stage. On that basis, multi-angle predicting models of wheat powdery mildew will be built and optimized using linear and nonlinear method, and then, the model will be validated and expanded through multi-angle imagery from nearground remote sensing platform. Above all, monitoring methods of wheat powdery mildew will been improved and multi-angle observation will be provided to monitor crop disease at large scale.

英文关键词： Hyperspectral imaging；Wheat Powdery mildew；Multi-angle；Feature extraction；Model