基于近地面高光谱影像特征与物联网架构的松材线虫病早期智能监测研究

Hyperspectral remote sensing with its powerful spectral sensitivity provides possibility for the early monitoring on forest diseases and pests. This research proposes to conduct continuous measurements of ground hyperspectral images by the fixed location and fixed plants experiments: the pine wood nematode inoculation, pine shoot blight bacteria inoculation and drought stress experiments, to obtain hyperspectral imaging features in different disease or stress periods and to resolve the problems such as: if the hyperspectral imaging features can realize the early distinguish among different pine species and different stress types or not? if the pine wilt disease degree can be quantified by hyperspectral imageing features or not? Based on the solution of these problems, the specific hyperspectral image bands or bands combination for early monitoring on the pine wilt disease will be used to produce the portable sensor with imaging, GPS location and communication function. This portable sensor can be used by the forest rangers to simply take some photos on the pine plants. These special photos will be immediately transmitted to the control center by wireless transmission network and then be analyzed by models in the software platform of the intelligent monitoring system to confirm the pine plants infected by pine wilt disease or not and to quantify the infection degree if its infection are confirmed. The confirm information will turn back to the portable sensor to instruct the forest rangers managing these pine plants in time. So, a portable intelligent system for early pine wilt disease monitoring based on the internet of things framework will be established. Successful completion of this proposed research will provide a method to meet the practical application demand of real-time monitoring function, effectively improve the prevention efficiency of forest pine wilt disease and provide the possibility of the scientific decision-making and rapid response to the pine wilt disease.

高光谱遥感以其强大的光谱敏感性为森林病虫害早期监测预警提供了可能。本研究拟通过定点定株松材线虫接种、松枯梢菌接种及干旱胁迫实验进行近地面连续高光谱影像测量，提取不同感病时期影像特征，解决不同松树种不同胁迫类型能否根据高光谱影像特征实现早期判别以及松材线虫病感病程度的定量化预测问题；在此基础上提取松材线虫病早期诊断特征影像波段制作成具备成像功能的便携传感器，并配置GPS定位与通迅功能，使巡林员通过简单的拍照操作，相应的携带定位信息的专用诊断波段影像即时通过无线传输网络传至控制中心，通过智能监测系统软件平台进行模型分析与诊断，确认林株是否感染松材线虫病及感病程度等信息，并将此信息即时传回给巡林员，其实时采取相应的管理调控措施。如此构建起基于物联网架构的便携式松材线虫病早期智能监测系统，将实现符合实际应用需求的实时监测功能，有效地提高森林松材线虫病的防治效率，为进行科学决策和快速反应提供了可能。

高光谱遥感以其强大的光谱敏感性能够探测到植被在病虫害侵害早期与健康植被的细微光谱差异，为森林病虫害的早期监测预警提供了可能。2015-2018年，按照研究计划，项目组连续四年进行了定点定株松材线虫病接种实验、干旱胁迫实验及同期松枯梢菌接种实验，获取连续完整的高光谱影像数据，通过对不同胁迫类型特征的动态分析与系统研究，项目组取得了具有较强准确性和普适性的研究结果。发表学术论文10篇，其中SCI收录2篇，软件著作权登记申请2项，培养研究生9名。分别建立近地面高光谱影像背景去除与辐射校正方法一套，建立黑松与马尾松松材线虫病早期监测方法两套，研制松材线虫病专用诊断相机（手持机载两用）一套，构建基于物联网架构的松材线虫病监测系统软件两套（单机版与网络版），构建诊断相机无人机搭载平台控制系统一套。研究获取了一系列以黑松与马尾松为代表的松材线虫病区别于其它胁迫类型的高光谱特征诊断指数，深入分析了松材线虫病早期（外观无变化）的高光谱影像特征与相应的生理参数变化特征的密切关系，对松树松材线虫病的动态监测实现了实时诊断、动态分析与定量描述的完整过程。同时，通过对同期不同胁迫类型的对比分析，获取干旱及松树其它病害的高光谱监测方法，为实现森林病虫害数字化监测平台奠定了坚实的理论基础与方法借鉴。项目成果经过完整地后期测试与实地验证，可进行产品化设计与包装后投入到森林健康管理与可持续经营实践中，具有极好的应用价值。与受基金资助，项目组开展了积极有效的学术交流。总体上，项目组较圆满地完成了既定的研究任务。