Construcción de UAV para la adquisición de imágenes multiespectrales en cultivos de quinua y su procesamiento con visión artificial.

Abstract

An unmanned aerial vehicle (UAV) was built for the acquisition of multispectral images in quinoa crops and their processing with artificial vision. The system is made up of a UAV that carries a multispectral camera as main equipment, which meets the requirements set out in this graduation work. To make the image acquisition flights possible, the technological methods were taken into account based on precision agriculture. Engine thrust calculations were essential, propellers and batteries for autonomous UAV flight. Verification of the connections of Communication between the UAV, the remote control and the telemetry module was done through the Mission Planner software, as well as the calibration of the flight control system and flight sensors of the Pixhawk controller. For the artificial vision system, images acquired with a RedEdge-MX brand multispectral camera were used, which provides 5 band samples for each shot. This camera uses a recommended height for taking images of 50 meters altitude in straight line. With the help of Matlab engineering software the near infrared bands were used (NIR) and visible red light (RED) to obtain the normalized difference vegetation index (NDVI), to which filtering, and segmentation processes were applied which were used to perform a comparison between samples with the presence of pests, in this case aphids and healthy plants, giving as a result an NDVI interval of 0.26 to 0.3 corresponding to the pest. As these data are verified by an agronomist engineer, it is concluded that the prototype is effective for the detection of organisms harmful to the plantation. It is recommended to extend the study topic to different crops and pests in order to have a wide field fof application.