Mazza, Antonio (2021) Deep Learning based data-fusion methods for remote sensing applications. [Tesi di dottorato]
Anteprima |
Testo
Mazza_Antonio_33.pdf Download (44MB) | Anteprima |
| Tipologia del documento: | Tesi di dottorato |
|---|---|
| Lingua: | English |
| Titolo: | Deep Learning based data-fusion methods for remote sensing applications |
| Autori: | Autore Email Mazza, Antonio antonio.mazza@unina.it |
| Data: | 29 Luglio 2021 |
| Numero di pagine: | 142 |
| Istituzione: | Università degli Studi di Napoli Federico II |
| Dipartimento: | Ingegneria Elettrica e delle Tecnologie dell'Informazione |
| Dottorato: | Information technology and electrical engineering |
| Ciclo di dottorato: | 33 |
| Coordinatore del Corso di dottorato: | nome email Riccio, Daniele daniele.riccio@unina.it |
| Tutor: | nome email Scarpa, Giuseppe [non definito] |
| Data: | 29 Luglio 2021 |
| Numero di pagine: | 142 |
| Parole chiave: | Deep Learning; Data-fusion; Remote sensing |
| Settori scientifico-disciplinari del MIUR: | Area 09 - Ingegneria industriale e dell'informazione > ING-INF/03 - Telecomunicazioni |
| Depositato il: | 27 Lug 2021 15:27 |
| Ultima modifica: | 07 Giu 2023 11:20 |
| URI: | http://www.fedoa.unina.it/id/eprint/13552 |
Abstract
In the last years, an increasing number of remote sensing sensors have been launched to orbit around the Earth, with a continuously growing production of massive data, that are useful for a large number of monitoring applications, especially for the monitoring task. Despite modern optical sensors provide rich spectral information about Earth's surface, at very high resolution, they are weather-sensitive. On the other hand, SAR images are always available also in presence of clouds and are almost weather-insensitive, as well as daynight available, but they do not provide a rich spectral information and are severely affected by speckle "noise" that make difficult the information extraction. For the above reasons it is worth and challenging to fuse data provided by different sources and/or acquired at different times, in order to leverage on their diversity and complementarity to retrieve the target information. Motivated by the success of the employment of Deep Learning methods in many image processing tasks, in this thesis it has been faced different typical remote sensing data-fusion problems by means of suitably designed Convolutional Neural Networks.
Downloads
Downloads per month over past year
Actions (login required)
 |
Modifica documento |
