Publication: Diseño y desarrollo de una aplicación para detección de cultivos mediante texturas
Loading...
Official URL
Full text at PDC
Publication Date
2017
Advisors (or tutors)
Montalvo Martínez, Martín
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
En el presente trabajo se diseña y desarrolla un sistema para la identificación del fruto en los cultivos agrícolas de calabacín. Para la identificación del fruto se emplean imágenes obtenidas por una cámara dentro del espectro visible en entornos de exterior. La principal complejidad radica en la diferenciación del fruto de las hojas y el tallo, ya que estos tres elementos presentan texturas semejantes que en ocasiones pueden ser complicadas de diferenciar hasta por un experto humano. El problema se ha abordado mediante el análisis de texturas obtenidas mediante la matriz de coocurrencia y el histograma. Con la información obtenida mediante la matriz de coocurrencia se entrena un sistema basado en aprendizaje supervisado y un sistema capaz de identificar el fruto del calabacín usando la distancia Euclídea y la de Mahalanobis.
In the present work a system for the identification of the product in the agricultural crops of zucchini is designed and developed. For the identification of the product we use images obtained by a camera within the visible spectrum in outdoor environments. The main complexity lies in the differentiation of the product of the leaves and the stem, since these three elements present similar textures that at times can be complicated to differentiate even by a human expert. The problem has been addressed by the analysis of textures obtained by the co-occurrence matrix and the histogram. With the information obtained through the cooccurrence matrix, a system based on supervised learning and a system is obtained able to identify the product of the zucchini using Euclidean distance and Mahalanobis distance.
In the present work a system for the identification of the product in the agricultural crops of zucchini is designed and developed. For the identification of the product we use images obtained by a camera within the visible spectrum in outdoor environments. The main complexity lies in the differentiation of the product of the leaves and the stem, since these three elements present similar textures that at times can be complicated to differentiate even by a human expert. The problem has been addressed by the analysis of textures obtained by the co-occurrence matrix and the histogram. With the information obtained through the cooccurrence matrix, a system based on supervised learning and a system is obtained able to identify the product of the zucchini using Euclidean distance and Mahalanobis distance.
Description
Trabajo de Fin de Grado en Ingeniería Informática (Universidad Complutense, Facultad de Informática, curso 2016/2017)