Name
Title | ||
|---|---|---|
Spectral monitoring of wheat canopy under uncontrolled conditions for decision making purposes. |
Abstract | ||
|---|---|---|
High quality of the crop multispectral images taken under uncontrolled conditions.High correspondence between air and ground spectral measurements.Quantifying the effect of lack of knowledge about the bidirectional reflectance distribution function.Considerations for the incorporation of the images in the decision making process. The increase in the supply of devices for acquiring spectral images has enabled its widespread adoption by farmers. These devices combine technical quality and ease of use, but the abandonment of experimental stations has led to the loss of the scientific supervision, which is necessary for the fulfilment of operating conditions. The aim of this study is to quantify the quality of the spectral images taken under uncontrolled conditions. With this objective, workflows in real plots were simulated for three years. Surveys were carried out by a spectral camera mounted on a ground platform without additional systems to control lighting, sun-target-sensor geometry and interferences. Images were processed using the empirical linear method and a set of five references, so that the goodness of fit is the first quality estimator. Over 97% of images reach a coefficient of determination equal to or above 0.75 at all wavelengths. The images were also compared with the spectral signatures obtained by a field radiometer and with images acquired by an airborne hyperspectral sensor. This allows us to quantify the error in determining the crop reflectance and to assess the effect of the uncontrolled conditions. The median error in comparison with data from the field radiometer is 0.01, 0.02 and 0.03 (green, red and near-infrared respectively) and with the airborne hyperspectral sensor the error is 0.01 or lower. The geometry is the key factor but it can be controlled and it is possible to use successfully the current methodologies for the agronomic interpretation. The monitoring quality is comparable, from a practical point of view, with more sophisticated instruments, but there is a need to find a solution to a bad timely operation of the camera for efficient operation of the workflow. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1016/j.compag.2016.05.002 | Computers and Electronics in Agriculture |
Keywords | Field | DocType |
BRDF,ELM,NDVI,PTFE,R2 | Bidirectional reflectance distribution function,Computer vision,Remote sensing,Multispectral image,Hyperspectral imaging,Normalized Difference Vegetation Index,Artificial intelligence,Engineering,Goodness of fit,Spectral signature,Radiometer,Estimator | Journal |
Volume | Issue | ISSN |
125 | C | 0168-1699 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 | ||