Application of a low-cost camera on a UAV to estimate maize nitrogen-related variables

被引:0
作者
Martina Corti
Daniele Cavalli
Giovanni Cabassi
Antonio Vigoni
Luigi Degano
Pietro Marino Gallina
机构
[1] Agroenergy,Department of Agricultural and Environmental Sciences – Production Landscape
[2] Università degli Studi di Milano,undefined
[3] Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria,undefined
[4] CREA-ZA,undefined
[5] Sport Turf Consulting-Servizi per l’agricoltura con aeromobili a pilotaggio remoto,undefined
来源
Precision Agriculture | 2019年 / 20卷
关键词
CIR camera; UAV; Colorgrams; Vegetation indices; Maize;
D O I
暂无
中图分类号
学科分类号
摘要
The development of small unmanned aerial vehicles and advances in sensor technology have made consumer digital cameras suitable for the remote sensing of vegetation. In this context, monitoring the in-field variability of maize (Zea mays L.), characterized by high nitrogen fertilization rates, with a low-cost color-infrared airborne system could be the basis for a site-specific nitrogen (N) fertilization support system. An experimental field with different N treatments applied to silage maize was monitored during the years 2014 and 2015. Images of the field and reference destructive measurements of above ground biomass, its N concentration and N uptake were taken at V6 and V9 development stages. Classical normalized difference vegetation indices (NDVI) and the indices adjusted by crop ground cover were calculated and regressed against the measured variables. Finally, image colorgrams were used to explore the potential of band-related information in variable estimation. A colorgram is a linear signal that summarizes the color content of each digital image. It is composed of a sequence of the frequency distribution curves of the camera bands, of their related parameters and of results of the principal components analysis applied to each image. The best predictors were found to be the ground cover and the adjusted green-based NDVI: regression equation at V9 resulted in R2 of 0.7 and RRMSE < 25% in external validation. Colorgrams did not improve prediction performance due to the spectral limitations of the camera. Therefore, the feasibility of the method should be tested in future research. In spite of limitations of sensor setup, the modified camera was able to estimate maize biomass due to the very high spatial resolution. Since the above ground biomass is a robust proxy of N status, the modified camera could be a promising tool for a low-cost N fertilization support system.
引用
收藏
页码:675 / 696
页数:21
相关论文
empty
未找到相关数据