Diseño e implementación de un sistema de visión artificial para análisis de datos NDVI en imágenes espectrales de cultivos de brócoli obtenidos mediante una aeronave pilotada remotamente.

Abstract

The aim of this research was to give an applied vision technique for the detection of characteristics in the evolution of broccoli growing. For the development of the system, a Survey 2 camera was included, which has a double band filter that reflects the red light to the red channel of the red, green, blue (RGB) sensor, and reflects the nearby light into the blue channel of the RGB sensor. The camera was adapted to an unmanned aerial vehicle and for its safety, coupled by a structure that was designed by the SolidWorks program and built in a 3D printer. The software design algorithm was made in the programming language Matlab, which consists of the digital processing of the spectral image, which bases its operation on the fact that the pigment in the leaves in the plant (chlorophyll) strongly absorbs the visible light for its use in photosynthesis and the cellular structure of the leaves, on the other hand, strongly reflects the light of the nearby infrared. The more leaves a plant has, the more wavelengths of light are affected respectively, that is why the purpose was to create a relation between the spectral image taken at a suitable height and the object to be studied (broccoli). Once the photographs were acquired with the camera, they were downloaded to a computer to be processed, obtaining the normalized differentiated vegetation index (NDVI), allowing to determine the quantity, quality and growth of the plants. The system has been tested in different broccoli crops in the province of Cotopaxi, bringing the expected results. It is recommended to expand the research with the development of software capable of automatically detecting areas with depressed vegetation througt the use of spectral indicators.