Publication: Técnicas de mejora de calidad del color en imágenes de exterior mediante comparación de histogramas
Loading...
Official URL
Full text at PDC
Publication Date
2014
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
En el presente trabajo se proponen dos métodos para la corrección del color en imágenes de exterior con el fin de mejorar su calidad.
Los métodos propuestos emplean el mismo procedimiento general: se selecciona una imagen de referencia con suficiente calidad, además se selecciona otra imagen de forma aleatoria del conjunto de imágenes a corregir, cuya calidad se considera insuficiente. A continuación se crea una tabla de correspondencias que establece una relación entre los histogramas de ambas, de forma que dicha tabla se puede emplear posteriormente para corregir cualquier imagen de la misma naturaleza que la de calidad insuficiente.
Para comparar la eficiencia emplearemos un tercer método que se diferencia de los anteriores en el hecho de que está basado en un algoritmo devorador, esto es basado en optimización local, con el que se obtienen los resultados pertinentes.
Se realiza un análisis comparativo sobre los resultados obtenidos por los tres métodos, no sólo en lo relativo a la calidad de la imagen resultante sino también en el tiempo de procesamiento, dado que se pretende que el método sea válido para procesamientos en tiempo real.
Los resultados obtenidos determinan que el método devorador obtiene resultados satisfactorios y que los otros métodos no lo mejoran, salvo casos puntuales. Se concluye presentando una serie de mejoras y alternativas para lograr resultados satisfactorios en los dos métodos propuestos.
In this work, we propose two methods for color correction in outdoor images in order to improve their quality. The proposed methods employ the same general procedure: a reference image with sufficient quality is selected; in addition another image is randomly chosen from the set of images available to be corrected, whose quality is considered insufficient. We create a table of correspondences that establishes a relationship between the histograms of both images, so we can use it later to correct any image of the same nature that the one with insufficient quality. To compare the efficiency of the proposed approaches we will use a third method that differs from the previous ones in that is based on a greedy algorithm, this is based on optimization. A comparative analysis of the results obtained by the three methods is realized, not only in regards to the quality of the resulting image but also to the processing time, given that it is intended that the proposed methods must fulfill with the requirements of real time processing. The results determine that the greedy method obtains satisfactory results and that the other methods do not improve them, except for specific cases. We conclude reporting some improvements and alternatives to achieve satisfactory results with both methods.
In this work, we propose two methods for color correction in outdoor images in order to improve their quality. The proposed methods employ the same general procedure: a reference image with sufficient quality is selected; in addition another image is randomly chosen from the set of images available to be corrected, whose quality is considered insufficient. We create a table of correspondences that establishes a relationship between the histograms of both images, so we can use it later to correct any image of the same nature that the one with insufficient quality. To compare the efficiency of the proposed approaches we will use a third method that differs from the previous ones in that is based on a greedy algorithm, this is based on optimization. A comparative analysis of the results obtained by the three methods is realized, not only in regards to the quality of the resulting image but also to the processing time, given that it is intended that the proposed methods must fulfill with the requirements of real time processing. The results determine that the greedy method obtains satisfactory results and that the other methods do not improve them, except for specific cases. We conclude reporting some improvements and alternatives to achieve satisfactory results with both methods.
Description
Máster en Investigación en Informática, Facultad de Informática, Departamento de Arquitectura de Computadores y Automática, curso 2013-2014