Multi-resolution spatial unmixing for meris and landsat image fusion
- Autores: Julia Amorós López
- Directores de la Tesis: Luis Gómez Chova (dir. tes.), Javier Calpe Maravilla (codir. tes.)
- Lectura: En la Universitat de València ( España ) en 2012
- Idioma: español
- Tribunal Calificador de la Tesis: Gustavo Camps Valls (presid.), José Moreno Méndez (secret.), Raúl Zurita Milla (voc.), Luis Guanter Palomar (voc.), Antonio Plaza (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TESEO
- Resumen
Earth observation satellites monitor our planet by acquiring images at different wavelengths of the electromagnetic spectrum. Exploitation of these images provides a better characterization of the different phenomena and changes that occur on Earth at local and global scales. During the last decades, remote sensing (RS) has been a very active field for both innovative science and operational applications, which have been mainly motivated by the increasing number of satellite missions as well as technological advances, processing tools, and data availability.
Nowadays, monitoring of the Earth's surface by RS satellites provides support to many public administrations and private companies. For instance, mapping agencies require accurate up-to-date land-cover maps and change detection indicators to enable better understanding of RS products and improved resource management, inventorying, and policy making. Furthermore, there is a need of data availability at high spatial resolution with temporal resolution shorter than one week.
However, technological constraints impose a trade-off between spatial and spectral resolutions, and between spatial resolution and coverage, i.e., high spatial resolution usually implies low spectral and temporal resolutions and vice versa. In this context, image fusion methods constitute a compelling field of research because they allow combining information from multiple sensors with different spatial, spectral, and temporal resolutions in order to obtain image products with improved overall characteristics.
This Thesis proposes a novel image fusion approach based on multiresolution and multisource regularized spatial unmixing. The proposed methodology yields a composite image with the spatial resolution of the higher spatial resolution image while retaining the spectral and temporal characteristics of the medium spatial resolution image. The approach is presented and tested for images from ENVISAT/MERIS and Landsat/TM instruments, but is general enough to be applied to other sensor pairs, such as the multispectral instruments flying onboard the ESA GMES Sentinel-2 and Sentinel-3 upcoming satellite series.
The potential of the methodology is illustrated in an agricultural monitoring application. The temporal profiles from the Landsat image series are complemented with results from the downscaled MERIS images, which allows a more accurate determination of the crop type and phenology as well as capturing rapidly varying land cover changes. The proposed methodology has been successfully illustrated suggesting that it may be very useful to cope with the immediate and future problems of multitemporal monitoring in remote sensing.
