Descripción
|
|
---|---|
A huge number of image fusion methods have appeared since the irruption of wavelets during last decades. New tools for multiscale geometric analysis (MGA) have been developed to improve the directional sensitivity. In this context appeared the Shearlet transform that is able to capture directional information and is equipped with a rich mathematical structure associated to multiresolution analysis. This transform has already shown its suitability to provide excellent results in image processing problems as denoising, image enhancement, image separation, edge detection and estimation of geometric features. The aim of this study is double: First to check the advantages of this transform for fusion comparing with Wavelet transform, and second to propose an efficient multidirectional & multiresolution algorithm to fuse panchromatic (PAN) and multispectral (MS) satellite images for SEOSAT, a Spanish satellite that will be launched in a next years. Four different methods are compared using Wavelet á trous transform (WAT) and Non-Subsampled Shearlet Transform (NSST). Results show that for simple fusion algorithms Shearlet Transform improves the Wavelet one but this is not true in general when comparing with a more complex fusion scheme like WAT weighted with Fractal Dimension Maps. Besides, a Shearlet based algorithm weighted with initial correlation provides as good results as more complex algorithms with a considerable reduction of computation time. Keywords. Shearlet transform, fusion image, multidirectional, multiresolution | |
Internacional
|
Si |
DOI
|
|
Edición del Libro
|
1 |
Editorial del Libro
|
IOS PRESS |
ISBN
|
978-88-89693-34-6 |
Serie
|
European Association of Remote Sensing Laboratories Proceedings |
Título del Libro
|
Towards Horizon 2020: Earth Observation and Social Perspectives |
Desde página
|
761 |
Hasta página
|
772 |