Questo sito utilizza cookie di terze parti per inviarti pubblicità in linea con le tue preferenze. Se vuoi saperne di più clicca QUI 
Chiudendo questo banner, scorrendo questa pagina, cliccando su un link o proseguendo la navigazione in altra maniera, acconsenti all'uso dei cookie. OK

Validazione sperimentale di metodologie di classificazione per immagini SAR multifrequenza e polarimetriche

Image classification is an important aspect in different applications. Specifically, image classification of remote sensing data is important for several applications like topography, oceanography, forestry, agriculture, urban planning, environmental and prediction and evaluation of natural disasters. This thesis is focused on image classification of SAR (Synthetic Aperture Radar) Images. A theoretical study is presented which summarises the most important aspects of radar and SAR and the theory about stochastic processes. We analysed SAR images with two different types of algorithms: one which is based on MRFs (Markov Random Fields) with ICM (Iterated Conditional Mode) that is a supervised classifier; the other based on K-Means algorithm. We combined these algorithms with different feature transformations and filters to obtain a better classification. To compare the results we used a statistical analysis based on “Confusion Matrices”. Particularly, our attention was dedicated to develop and test the MRF-ICM method for classification of SAR images. In the past, these methods are not used for classification of only SAR images but also with optical images. In addition, we studied the image properties in order to obtain the better data representation. Among these properties, we consider feature transformations in order to study the image from a different points of view. In addition, we tested the accuracy of the Log-normal probability density function to describe the image and a relative classification. Instead, for the unsupervised classification, we used, as described above, K-Means algorithm which was used to compare different classifications. Actually, we studied the feature extraction of PolSAR (Polarized SAR) images which consists in parameters alpha and entropy (α and H). Thus, the results consist in classifications with these parameters and, then, we compare the results of supervised and unsupervised classification separately.

Mostra/Nascondi contenuto.
INTRODUCTION 1 INTRODUCTION Environmental monitoring, earth-resource mapping, and defence (security systems) may involve broad-area imaging at high and medium resolution. The imagery must often be acquired in inclement weather or during night as well as day. Synthetic Aperture Radar (SAR), both on airborne and spaceborne platforms, provides such a capability. SAR systems take advantage of the long-range propagation characteristics of radar signals and of the complex information processing capability of modern digital electronics to provide high and median resolution imagery. Synthetic Aperture Radar complements photographic and other optical imaging capabilities thanks to the minimum constraints on time-of-day and atmospheric conditions and thanks to the unique responses of terrain and cultural targets to radar frequencies. ERS-1 was the first instrument in a series of orbital SAR planned to have long lifetimes and semi-operational capabilities. Consequently ERS-2, JERS-2 and RADARSAT satellite systems were created. ENVISAT was launched in 2001. These satellites allow dynamic processes to be observed over most of the Earth’s surface by providing a long series of accurate measurements of the backscatter coefficient 1 . This has a significant impact in many scientific domains: vegetation mapping and monitoring, hydrology, sea-ice mapping and geology. Particularly, Synthetic Aperture Radar technology has provided terrain structural information to geologists for mineral exploration, oil spill boundaries on water to environmentalists, sea state and ice hazard maps to navigators, and reconnaissance and targeting information to military operations. There are many other applications or potential applications. Some of these, particularly civilian, have not yet been adequately explored because lower cost electronics are just beginning to make SAR technology economical for smaller scale uses. Spaceborne radar systems use single frequencies and polarization with modest resolution, because of the constraints imposed by their deployment in space. On the contrary, more complex airborne systems have demonstrated the advantages of multiple frequencies and polarizations. These advantages 1 The backscattering or backward scattering coefficient, in unit of m -1 , indicates the attenuation (reduction in light intensity) caused by scattering at angles from 90° to 180°. 1

Laurea liv.I

Facoltà: Ingegneria

Autore: Alessandro Carrega Contatta »

Composta da 158 pagine.

 

Questa tesi ha raggiunto 2397 click dal 23/05/2006.

 

Consultata integralmente 2 volte.

Disponibile in PDF, la consultazione è esclusivamente in formato digitale.