Desarrollo de un algoritmo que determine los factores que afectan a la calidad de los cultivos mediante procesamiento de imágenes espectrales.

Abstract

The project addresses a critical problem in Ecuadorian agriculture: the lack of algorithms to identify factors that affect crop quality. The goal was to develop an accurate algorithm that uses multispectral imaging remote sensing to assess the factors that affect crops, in this case, maize. In terms of methodology, the study was carried out in a corn field in the province of Chimborazo, using specialized technology. A Micasense Rededge Mx multispectral camera and a Phantom 3 drone were used. To maximize detail, images were captured at moderately low altitudes, no higher than 35 meters. In addition, the camera was attached to the drone using a device designed in SolidWorks and printed in 3D, which allowed for more stable capture and minimized distortions caused by wind. Regarding the results, the algorithm processes the images with various techniques such as radiometric and geometric calibration, and the conversion from radiance to reflectance. The results highlight variations in the intensity of the color that reveal the state of health of the crop and the affecting factors. Chlorophyll and humidity levels were also evaluated through vegetation indices. In conclusion, the algorithm proved to be effective in the identification and quantification of factors that significantly affect the quality of corn. Its implementation could allow farmers and experts to make more informed decisions, contributing to the optimization and development of the agricultural sector in Ecuador.