基于无人机图像的玉米冠层叶绿素含量检测与分布研究

Chlorophyll Content Detection and Distribution Research of Maize Canopy Based on UAV Image

为了快速、无损地获取大田作物叶绿素含量空间分布,基于无人机遥感技术研究了大田玉米冠层叶绿素含量检测及分布图绘制方法。利用无人机遥感技术采集了150幅大田玉米的航拍图像,并通过Pix4dmapper软件对其进行了拼接;在实验田中,等距获取80株玉米叶片样本,通过化学法萃取叶绿素,并使用分光光度计测量叶绿素含量,形成了基础数据源。在数据处理方面,采用ArcGIS软件对样本点的POS(Position and orientation system)数据与无人机图像进行匹配;对无人机拍摄的RGB图像,首先进行 R、G、B 三通道分量值提取,构建了绿红比值、绿红差值、归一化红绿差值、超绿等10种颜色特征,并计算了均值、标准偏差、平滑度、三阶矩等6种纹理特征,然后建立了基于BP(Back propagation)神经网络的玉米冠层叶绿素含量检测模型。实验结果表明,基于BP神经网络的玉米冠层叶绿素含量检测模型的均方根误差RMSE为4.465 9 mg/L,决定系数 R 2 为0.724 6。通过BP神经网络检测模型计算出大田玉米图像每个像素点的叶绿素含量,基于伪彩色技术绘制大田玉米叶绿素含量可视化分布图,分析田间玉米冠层叶绿素含量分布图可以直观区分田间道路与冠层区域,显示地块叶绿素分布差异。通过无损检测大田玉米冠层叶绿素含量及叶绿素分布可视化,可为田间作物长势评价和精细化管理提供技术支持。

Chlorophyll is an important indicator for the evaluation of plant photosynthesis ability and growth status.In order to obtain the spatial distribution of chlorophyll content in field crops quickly and non-destructively,the chlorophyll content detection and distribution map drawing method of maize canopy were carried out based on UAV remote sensing technology.Firstly,the aerial images of 150 maize plots were collected by UAV mounted camera and spliced by Pix4dmapper software.Totally 80 maize leaves were sampled in the experimental field.They were processed following chemical extraction and spectrophotometer measurement to obtain the chlorophyll content value.The images and chlorophyll data were used to form the underlying data source.In the aspects of data processing,the position and orientation system (POS) data of the sample points were matched with the images of the UAV using ArcGIS software.For the RGB images captured by the drone,the three-channel component values of R,G and B were firstly extracted.The color feature parameters were calculated such as green-red difference,normalized red-green difference,super green,and so on.In addition,six kinds of texture features were calculated,including mean,standard deviation,smoothness and third-order moment.The error back propagation neural network was used to build chlorophyll detection model for maize canopy leaves.The experimental results were as follows: the root mean square error (RMSE) of the maize canopy chlorophyll content detecting model based on BP neural network was 4.465 9 mg/L,and the coefficient of determination R 2 was 0.724 6.The chlorophyll content of each pixel in the field canopy image was calculated.The visual distribution map of chlorophyll content in field maize canopy was drawn based on pseudo-color technique.The chlorophyll content distribution map of field maize canopy could be used to visually distinguish the field road and canopy area,showing the difference in chlorophyll distribution of the plot.By non-destructively detecting the chlorophyll content and chlorophyll distribution of canopy corn canopy,it could provide a support for field crop growth evaluation and precision management.