Memorias de investigación
Ponencias en congresos:
Toward multi-scale object-based images fusion
Año:2011
Áreas de investigación
  • Ingenierías,
  • Procesamiento de imágenes
Datos
Descripción
oday, a plethora of image fusion algorithms are available, most of them belong to the pixel-by-pixel level paradigm. They present the handicap that is impossible to represent the whole visual information contained in a set of images as a single fused image without loss of information. A common characteristic to all these fusion strategies is that they do not distinguish between different land cover areas; that means that the same information extracted from the panchromatic image is integrated in the whole spectral band, independently of the surfaces characteristics present in it. An important negative consequence of these techniques is that introduce false spatial details in fused images, reducing its usefulness in typical remote sensing tasks such as image classification and change detection. This justifies the need of new fusion strategies which adapt the fusion process, not only to each one of the spectral image bands, but also to each one of the type-coverage presented in it. These strategies would allow differentiating the quantity and quality information coming from each of source images that should be integrated in the final fused image. In this sense, the aim of this work is to develop a new strategies of satellite images fusion objects-based, for agricultural applications. The first step of the methodology corresponds to land cover characterization of source images at different spatial-scales. In this sense, morphological image processing will be applied for isolate and characterize the objects of interest present in the image. Logical operators, like dilatation, erosion and smoothing and their combinations (opening and closing) will be used for that. Local lacunarity analysis (applied to gray level image) will be used to textural characterization of the land-cover. Thus, data sets with object attributes will be obtained. From this information, a debugged and association process should be carried out. Finally, these information will be processed using high-level fusion rules to obtain an object-based fused image.
Internacional
No
Nombre congreso
EARSEL 2011
Tipo de participación
960
Lugar del congreso
Praga (Chequia)
Revisores
Si
ISBN o ISSN
978-80-01-04868-9
DOI
Fecha inicio congreso
30/05/2011
Fecha fin congreso
02/06/2011
Desde la página
401
Hasta la página
406
Título de las actas
Remote Sensing and Geoinformation not only for Scientific Cooperation
Esta actividad pertenece a memorias de investigación
Participantes
Grupos de investigación, Departamentos, Centros e Institutos de I+D+i relacionados
  • Creador: Grupo de Investigación: Informática Aplicada al Procesado de Señal e Imagen