欧亚国际

   Airphen多光谱相机专门设计用于田间表型实验中，利用无人机搭载或其它动力机械搭载或固定或手持来评估测量冠层生物物理变量。相机仅重200g，可记录GPS和IMU信息，可激发控制伴载的热成像相机或高分辨率RGB相机。系统选用了6个窄波段来获取叶绿素特异系数，记录反射辐射情况，分辨率可达1280X960像素。8-mm焦距所提供的地面分辨率从几个mm到几个cm，这取决于飞行高度。6波段相机配备了4.2mm焦距镜头，用来简化图像对齐过程，生成较有效的D密度点云，通过该波段大测绘带实现高通量测量。我们开发了称之为“phenoscript”的通路来处理AIRPHEN相机的原图，计算输出为多光谱正射影像，从原图中提取微点以及对应的D密度点云。提取部分可之后可用于计算植被指数，测量冠层高度，基于经验方法或反转辐射传输模型获取数个生物物理变量。AIRPHEN多光谱相机的使用是在对一系列植物包括小麦、玉米、甜菜和土豆的几个实验基础上加以阐释的。植物指数动力学显示一段时间内的测量具有良好的一致性。我们就几个生物物理参数精度进行了预估，包括绿色覆盖率 (GF)，绿面积指数 (GAI)， 辐射截获率(FIPAR) ，冠层 (CCC) 或叶绿素含量 (LCC) 。在鉴别水胁迫的试验中，也对如何将多光谱相机以及热红外成像结合起来加以介绍。研究也探讨了AIRPHEN多光谱相机的开发使用，包括基于3D冠层机构来对图像进行先进解读，Airphen多光谱相机业可安装在其它装置如phenomobiles，较灵活机动，或者安装在手持系统上，专门用于小型和低成本实验。AIRPHEN也是田间物联网固定表型平台的基础。

AIRPHEN: A multispectral camera dedicated to field phenotyping from drone observations

A. Comar1, F. Baret2, G. Collombeau2, M. Hemmerlé1, B. de Solan3, D. Dutartre4, M. Weiss2, S. Madec2, F. Toromanoff3

1HIPHEN, Avignon, France

2INRA, UMR EMMAH, Avignon, France

3ARVALIS Institut du végétal, Avignon, France

4ITB, Avignon, France

The AIRPHEN multispectral camera was specifically designed for accessing canopy biophysical variables from drone observations within field phenotyping experiments. The camera weighs about 200 g, records GPS and IMU information and allows triggering companion devices including a thermal infrared or RGB high resolution camera. Six narrow wavebands selected to sample the chlorophyll specific absorption coefficient record the reflected radiation at a resolution of 1,280 x 960 pixels. The 8-mm focal length provides resolutions at ground level ranging from a few millimeters to a few centimeters, depending on flight altitude. One of the six cameras is equipped with a 4.2-mm focal length lens to ease the image alignment process and generate more accurate 3D point clouds, while providing the increased throughput allowed by the larger swath of the camera in this band. A pipeline called ‘phenoscript’ was developed to process the raw images of the AIRPHEN camera, compute as outputs a multispectral ortho-image, and extract the microplots from the original images and the corresponding 3D dense point cloud. The extracts can later be used to compute vegetation indices, measure canopy height, and derive a few biophysical variables from either empirical approaches or radiative transfer model inversion.

The use of the AIRPHEN camera is illustrated in several experiments conducted on a range of species including wheat, maize, sugar beet and potatoes. The dynamics of the vegetation indices show a very good consistency of the measurements over time. The accuracy of estimates of several biophysical variables including the green fraction (GF), the Green Area Index (GAI), the fraction of intercepted radiation (FIPAR) and canopy (CCC) or leaf (LCC) chlorophyll content is presented. The capacity to combine the multispectral camera with thermal infrared imagery is also illustrated by the characterization of water stress. Further developments on the use of the AIRPHEN camera are discussed, including advanced interpretation of the images based on 3D canopy structure models,installation on other vectors such as the phenomobile for working in active mode, or a handheld system dedicated to small and low-cost experiments.