Using UAVs and digital image processing to quantify areas of soil and vegetation

4th International Conference on Mathematical Modeling in Physical Sciences

物理学;数学

Chaves, A.A.^1,3 ; La Scalea, R.A.^2 ; Colturato, A.B.^1 ; Kawabata, C.L.O.^2 ; Furtado, E.L.^3 ; Branco, K.R.L.J. Castelo^1

Institute of Mathematical and Computer Sciences, University of Sao Paulo, Sao Carlos, Brazil^1

Federal Institute of Education, Science and Technology of Sao Paulo, Sao Carlos, Brazil^2

Faculty of Agronomical Sciences, State University of Sao Paulo, Botucatu, Brazil^3

关键词: Area measurement;    Crop monitoring;    Eucalyptus plantations;    Image processing algorithm;    Monitoring applications;    Other properties;    Planted areas;    Spatial resolution;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/633/1/012112/pdf
DOI  :  10.1088/1742-6596/633/1/012112

来源: IOP

【 摘 要 】

Unmanned aerial vehicles (UAVs) are becoming a very popular tool for remote sensing and crop monitoring. They are more easily deployed, cheaper and can obtain images with higher spatial-resolution than satellites. Some small, commercial UAVs can obtain images with spatial-resolution as low as 1.5cm per pixel. This opens up the range of possible remote sensing and monitoring applications. Moreover, they can cover large areas in very little time, such as 50 ha in about 20min, which makes UAVs the ideal tool for monitoring large farms and plantations. On the other hand, it is important to know precisely the area covered by farms in order to avoid invasion of other properties or preserved areas, and also to detect flaws in the plantation area. However, it is difficult to measure planted areas in some cases, such as Eucalyptus crops. Therefore, this paper aims to evaluate the use of UAV imagery for precise area measurement in Eucalyptus crops. We developed an image-processing algorithm to segment regions of soil, low biomass and high biomass and tested it on a Eucalyptus plantation in the city of Lenis Paulista -SP, Brazil. Results show that the area quantification is very accurate especially for bare soil regions and this method can be used to estimate areas in other scenarios.