李长春,牛庆林,杨贵军,冯海宽,刘建刚,王艳杰.基于无人机数码影像的大豆育种材料叶面积指数估测[J].农业机械学报,2017,48(8):147-158.
LI Changchun,NIU Qinglin,YANG Guijun,FENG Haikuan,LIU Jiangang,WANG Yanjie.Estimation of Leaf Area Index of Soybean Breeding Materials Based on UAV Digital Images[J].Transactions of the Chinese Society for Agricultural Machinery,2017,48(8):147-158.
摘要点击次数: 2504
全文下载次数: 1034
基于无人机数码影像的大豆育种材料叶面积指数估测   [下载全文]
Estimation of Leaf Area Index of Soybean Breeding Materials Based on UAV Digital Images   [Download Pdf][in English]
投稿时间:2017-05-08  
DOI:10.6041/j.issn.1000-1298.2017.08.016
中文关键词:  大豆育种材料  叶面积指数  标定  无人机  数码影像  全子集回归
基金项目:国家自然科学基金项目(41601346、 61661136003、41601364、41271345)、北京市农林科学院科技创新能力建设项目(KJCX20140417)和河南省基础与前沿研究项目(152300410098)
作者单位
李长春 河南理工大学 
牛庆林 河南理工大学
农业部农业遥感机理与定量遥感重点实验室 
杨贵军 农业部农业遥感机理与定量遥感重点实验室 
冯海宽 农业部农业遥感机理与定量遥感重点实验室 
刘建刚 农业部农业遥感机理与定量遥感重点实验室 
王艳杰 河南理工大学
农业部农业遥感机理与定量遥感重点实验室 
中文摘要:利用低成本的无人机(Unmanned aerial vehicle, UAV)高清数码影像获取系统,于2016年7—9月在山东省济宁市嘉祥县圣丰大豆育种基地,获取大豆育种材料试验区的R1-R2、R3、R5-R6共3个关键生育期的高清数码影像,首先利用高清数码影像中的黑白定标布,对数码影像的DN(Digital number,DN)值进行归一化标定,并构建标定的18个数码影像变量,然后基于900个育种小区的叶面积指数实测数据构建大豆育种材料叶面积指数的一元线性回归、逐步回归、全子集回归、偏最小二乘回归、支持向量机回归和随机森林回归模型,最后基于模型建立和验证的决定系数(R2)、均方根误差(RMSE)和归一化的均方根误差(nRMSE)3个指标,筛选估测叶面积指数的最佳模型。研究表明,全子集回归模型中采用4个数码影像变量B、RGBVI、GLA和B/(R+G+B)的多元线性回归模型对大豆育种材料叶面积指数的解析精度最优,模型建立的R2、RMSE和nRMSE分别为0.69、0.99和17.90%,验证模型的R2、RMSE和nRMSE分别为0.68、1.00和18.10%。结果表明,以无人机为遥感平台,搭载低成本的高清数码相机,利用高清数码影像进行大豆育种材料LAI估测是可行的,可以快速、有效、无损地获取大豆育种材料的长势信息,为筛选高产大豆品种提供一种低成本的可行方法。
LI Changchun  NIU Qinglin  YANG Guijun  FENG Haikuan  LIU Jiangang  WANG Yanjie
Henan Polytechnic University,Henan Polytechnic University;Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture,Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture,Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture,Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture and Henan Polytechnic University;Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture
Key Words:soybean breeding materials  leaf area index  calibration  unmanned aerial vehicle  digital images  total subsets regression
Abstract:Soybean is an important source of protein and fat. The increase of soybean yield is playing a significant role in guaranteeing food security and satisfying market demanding. Therefore, rapid screening of soybean varieties with high yield and quality is of great significance to increase the total output of soybean. Leaf area index (LAI), which refers to the gross one-sided leaf area per surface area, is one of the critical phenotypic parameters to characterize crop canopy structure, and it has an important significance to evaluate crop photosynthesis, growth and predict yield. A rapid, non-destructive and efficient estimation of soybean LAI can assist the screening of high-yield varieties. Currently, lots of soybean breeding material plots is one the difficulties in soybean breeding, but traditional manual investigation method is time-consuming, inefficient job with certain degree of subjectivity. Unmanned aerial vehicle (UAV) remote sensing technology has become a research focus on precision agriculture application. It features the advantages of easy construction, low operation and maintenance cost and flexible mobility, and has been used to realize rapid, non-destructive, spatial continuous crop growth monitoring and crop yield estimation. Researches based on low-cost UAV high spatial resolution digital images to estimate crop phenotypic parameters mainly focused on the crop cultivation and management sector. However, there are few researches on crop breeding. The high spatial resolution digital images of the Shengfeng academician workstation of soybean breeding experiment located in Jiaxiang County, Jining City, Shandong Province, China from July to September in 2016 were acquired using a low-cost UAV digital camera system. The obtained UAV data contained the high spatial resolution images of growth periods of R1-R2, R3 and R5-R6. At the same time, the average LAI values of 900 breeding plots on the ground were obtained. Firstly, the digital orthophoto map (DOM) was generated. The generated DOM was calibrated using the image values of black and white calibration tarps in the DOM image and a total of eighteen calibrated variables of R, G, B, MGRVI, RGBVI, GLA, ExG, WI, ExGR, CIVE, VARI, G/R, G/B, R/B, R/(R+G+B), G/(R+G+B) and B/(R+G+B) were calculated based on existing research. Secondly, 70% of the total data pairs of the eighteen variables and corresponding ground measured data were used to build models, including the unary linear regression, stepwise regression, total subset regression, partial least squares regression, support vector machine regression and random forest regression, while the remaining data pairs were used for model validation. Finally, the optimal prediction model for LAI was selected by comprehensively considering the determination coefficient (R2), root mean square error (RMSE) and normalized root mean square error (nRMSE) of model building and validating. The results showed that the total subset regression, which included four variables of B, RGBVI, GLA and B/(R+G+B), was the optimal estimation model of LAI. The R2, RMSE and nRMSE of model building and validation were 0.69, 0.99, 17.90% and 0.68, 1.00, 18.10%, respectively. The spatial distribution map of LAI of soybean breeding materials area was generated. Compared with ground-measured values and DOM derived from digital camera images, the distribution map could well reflect the growth status of soybean breeding materials. The results showed that high spatial resolution digital images of soybean breeding materials could be obtained quickly using UAV remote sensing technology. After that, the qualitative and quantitative analysis can be carried out to monitor the status of soybean breeding materials in the study area. In general, the UAV remote sensing technology with digital camera was feasible in predicting the LAI of soybean breeding materials, and it can serve as a rapid, effective and non-destructive way for LAI estimation in large-scale soybean breeding area.

Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.

   下载PDF阅读器