Descripción
|
|
---|---|
Remote sensing area is characterized by the use of large data volumes with a multidimensional nature, where graphic visualization of multispectral patterns becomes complicated. Visual data mining tries to enable users to explore massive high dimensional data sets to search interesting and useful information. Self-organizing map projects high-dimensional signal spaces on a two-dimensional displayable space, compressing information while preserving the most important topological relationships of the training patterns. These characteristics make self-organizing map a useful tool for multidimensional patterns visualization. In this work, a projection method of multidimensional data in two-dimensional space using growing self-organizing maps is proposed. With this projection technique, several traditional methods for visual analysis of Kohonen¿s self-organizing maps have been implemented for growing self-organizing maps. Different groups of multispectral data of images registered by Landsat-ETM+ and QuickBird satellites have been used to perform an exploratory visual analysis with this new graphical technique. | |
Internacional
|
Si |
Nombre congreso
|
IEEE International Geoscience and Remote Sensing Symposium |
Tipo de participación
|
960 |
Lugar del congreso
|
Barcelona |
Revisores
|
Si |
ISBN o ISSN
|
1-4244-1212-9 |
DOI
|
|
Fecha inicio congreso
|
23/07/2007 |
Fecha fin congreso
|
27/07/2007 |
Desde la página
|
|
Hasta la página
|
|
Título de las actas
|