Desarrollo de una metodología para estimar contenido de nitrógeno en maíz (Zea mays) mediante análisis de imágenes
Loading...
Date
2014
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
Se desarrolló una metodología para estimar el contenido de nitrógeno en plantas de maíz mediante el procesamiento de imágenes Esto es posible ya que el contenido de nitrógeno en las plantas se expresa en distintos niveles de intensidad de verde/amarillo. Las imágenes provenían de plantas con niveles de nitrógeno conocido, medidos mediante un instrumento denominado SPAD. A cada hoja se le determinó un valor medio de Unidades SPAD y un valor representativo de color, con el propósito de correlacionar sus valores. Para generar un valor representativo de color para cada hoja, los pixeles fueron agrupados en función de su grado de similitud empleando la técnica denominada K-means, técnica que fue aplicada sobre las imágenes en los espacios de color RGB, L*a*b* y HSV. Una vez agrupados los píxeles se creó un valor representativo de la hoja a partir de un promedio ponderado de los píxeles representantes de cada clase. Los resultados muestran que es posible correlacionar el color con el contenido de nitrógeno en los canales G, b* y V.
A methodology to estimate the nitrogen content in maize plants using image processing was developed. This is possible since the content of Nitrogen in plants is expressed at different levels of intensity of the colors green / yellow. The images came from plants with known levels of nitrogen, measured by an instrument called SPAD. In order to establish a correlation between both values, an average value of SPAD units and a representative color value were assigned to every leaf. To generate a representative color value for each leaf, pixels were grouped according to their degree of similarity using the k-means technique, which was applied to the images in the RGB, L*a*b * and HSV color space. Once pixels were grouped, a representative value of the leaf was created from a weighted average of the representative pixels of each group. Results show that it is possible to correlate colour with the content of nitrogen in the channels G, b * and V.
A methodology to estimate the nitrogen content in maize plants using image processing was developed. This is possible since the content of Nitrogen in plants is expressed at different levels of intensity of the colors green / yellow. The images came from plants with known levels of nitrogen, measured by an instrument called SPAD. In order to establish a correlation between both values, an average value of SPAD units and a representative color value were assigned to every leaf. To generate a representative color value for each leaf, pixels were grouped according to their degree of similarity using the k-means technique, which was applied to the images in the RGB, L*a*b * and HSV color space. Once pixels were grouped, a representative value of the leaf was created from a weighted average of the representative pixels of each group. Results show that it is possible to correlate colour with the content of nitrogen in the channels G, b * and V.
Description
Tesis presentada para optar al título de Ingeniero Civil Agrícola
Keywords
Maíz, Nitrógeno, Agricultura de precisión - Chile