USO DO SENSOR KINECT PARA MEDIR A REGULARIDADE NA DISTRIBUIÇÃO DE FERTILIZANTES SÓLIDOS

• Autor(a) principal: Ramos, Marcos Roberto de
• Data de Publicação: 2012
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UEPG
• Texto Completo: http://tede2.uepg.br/jspui/handle/prefix/158
• topic: Kinect; calibração; fertilizantes; infravermelho; fertisystem; Kinect; calibration; fertilizers; infrared; fertisystem; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

With the advent of computer vision, researchers have been investigating methods of capturing images of 3D objects and use this information in various fields such as medicine, architecture, education, among others. The launch of Xbox KinectTM sensor in November 2010 as an accessory to detect player movement, drew attention not only for its low cost, but also by providing information unencrypted, allowing their use for other tasks beyond videogames.This study explored the ability of the sensor to obtain distances of depth through its infrared camera and apply it to solving a problem in agronomy related to regularity in the distribution of fertilizers. It is known that irregularity in the distribution of fertilizers can cause serious problems to the crops such as excess or lack of nutrients, and lead to economic losses and environmental damage by the excess of nutrient applied. These problems are related to several factors, including the adjustments of distribution equipment or even the quality of fertilizers. To achieve the proposed goal, we developed a detailed study on the functioning of the sensor and test the accuracy of the calculations and measurements. Results showed 99.78% accuracy up to the detection area and volume occupied by the materials tested, with variations in tenths of a millimeter. A computer system was designed to obtain the fertilizer mass, calculating its volume using the sensor information knowing the density of each material. In validation experiments with fertilizers system reached up to 98.26% of accuracy with variations in tenths of grams of material estimated by the system and the actual mass measured by a precision balance. The system was tested on measurements of the mass of fertilizer on a belt in motion,simulating the application of a fertilizer rate. A hit above 98% from the estimated amount and the actual amount applied was obtained. The results suggest that it is feasible the use of KinectTM sensor for measuring the regularity in fertilizers distribution by using the methods and techniques experimented in this study.