DSpace - Tor Vergata >
Facoltà di Ingegneria >
Tesi di dottorato in ingegneria >

Please use this identifier to cite or link to this item: http://hdl.handle.net/2108/1076

Full metadata record

DC FieldValueLanguage
contributor.advisorSolimini, Domenico-
contributor.authorPutignano, Cosimo-
coverage.spatialColli Albanien
description.abstractThe polarimetric observables in a SAR image possess an intrinsic physical information, what makes polarimetric data fit to unsupervised classification, without need of a-priori information on the scene. Indeed, in natural targets, like vegetation, surface, volume and sometimes double-bounce scattering mechanisms are mixed, while backscattering from man-made targets can be usually attributed to dihedrons, trihedrons and bare surfaces. In many cases a radar resolution cell hosts more than one mechanism, although an average or dominant scattering mechanism can be identified for the purposes of classification. Following Chandrasekhar's pioneering target decomposition and the generalized and systematic theory by Huynen, a number of approaches to the interpretation of the scattering processes and to the identification of scatterers have appeared in the open literature. Target decomposition theory laid down the basis for the classification of radar images. In particular, the formalism worked out by Cloude, led to the introduction of an unsupervised classification scheme, further augmented and improved by subsequent contributions, also connecting the fuzzy logic theory with Wishart's statistical approach and electromagnetic modeling. Neural Network Algorithms (NNA) have been used in multispectral images classification and for change maps, but their application to polarimetric SAR image classification is more limited. In supervised schemes, the NNA were trained by Huynens parameters, or by the polarimetric coherence matrix [T], H and alpha. Unsupervised Neural Net classifiers, based on Self-Organizing Maps (SOM), have exploited Mϋller matrix directly, polarization signatures, or parameters derived from decomposition, like Huynen's, or Freeman's. In this Thesis two novel unsupervised classification algorithms, named PolSOM and TexSOM, for polarimetric data are proposed. Both algorithms are SOM-based and have been tested on complex Italian landscapes, where classification can become quite challenging and a limited use of polarimetric data has been reported for undulating, heterogeneous and fragmented scenarios. AIRSAR fully polarimetric data from MAC-Europe Campaign and RADARSAT-2 data acquired for a SOAR project (SOAR-1488) have been classified and confusion matrices have been computed from ground truth maps. PolSOM and TexSOM performances have been compared with each other and with consolidated and commonly used classification method, to assess their potential. The Neural Network algorithms have been carefully designed based on an in-depth analysis of their operation and, for the first time at the author's knowledge, both object-based and pixel-based information are jointly used in Radar polarimetric image analysis. The proposed classification algorithms are proving to be fairly versatile and not strictly confined to polarimetric images, like the other considered algorithms.en
description.tableofcontentsIntroduction - 1 State of the Art - 1.1 Radar polarimetry historical background - 1.2 Classification Techniques - 1.2.1 Target Decomposition - Coherent Decomposition Theorems - Mϋller matrix and Stokes vector Decomposition Theorem - TD based on the eigenvector analysis of the covariance or coherency matrix - 1.2.2 Unsupervised - 1.2.3 Supervised - 1.2.4 Neural-Network Based - Supervised - Unsupervised - Hybrid Neural Network algorithms - 1.2.5 Synopsis - 1.3 Motivations & Innovations of this research. - 2 Polarimetry and SOM background - 2.1 Polarimetry background - 2.1.1 Stokes Formalism - 2.1.2 Jones Formalism - Polarization Basis Change - 2.1.3 Complex polarization ratio - 2.1.4 Huynen's polarization fork - 2.1.5 Polarization Signatures and pedestal height - 2.1.6 Scattering Source Grouping - 2.1.7 Backscattering processes matrix analysis - 2.1.8 The Scattering Matrix - 2.1.9 The Mϋller Matrix - 2.1.10 Basis change for [S] Matrix - 2.1.11 Complex Scattering Vector - 2.1.12 Pauli Decomposition - 2.1.13 Non deterministic Scatterers - 2.1.14 Eigenvector-Eigenvalue based and H/A/alpha-Wishart decompositions - 2.1.15 Polarimetric classification preserving polarimetric scattering characteristics - 2.2 Self-organizing Maps background - 2.2.1 Introduction - 2.2.2 Neural Networks Model - 2.2.3 Mathematical Model - 2.2.4 SOM Structure - 2.2.5 Parameters Setting - Dimension - Learning Rate - Neighborhood Function - Training Cycles - 2.2.6 Learning Process - 2.2.7 SOM characteristics - Codification of input Space - Organization - Approximation of data density of probability - 2.2.8 SOM visualization: The U-matrix - 2.2.9 SOM Clustering - Hierarchical approach - K-Means approach - SOM based re-clustering approach. - 3 PolSOM and TexSOM in Polarimetric SAR Classification - 3.1 Introduction - 3.2 Data Set - 3.2.1 NASA/JPL AIRSAR - Montespertoli Test Site and Ground Truth - Montespertoli data overview - 3.2.2 RADARSAT-2 - Tor Vergata Colli Albani Test Site and Ground Truth - Tor Vergata Colli Albani data analysis - 3.3 PolSOM - 3.3.1 Introduction - 3.3.2 PolSOM IDL code developing - 3.3.3 Input data pre-processing - 3.3.4 New SOM Training technique - Supervised Training - Gaussian training data set- Mixed Training data set - Unsupervised Training - 3.3.5 AirSAR C-, L- and P-band polarimetric data classification with PolSOM - 3.3.6 AirSAR C-, L- and P-band polarimetric data H/A/alpha-W classification and comparison - 3.3.7 RADARSAT-2 PolSOM classification - 3.3.8 RADARSAT-2 classification preserving polarimetric scattering characteristics and comparison with PolSOM - 3.4 TexSOM - 3.4.1 Introduction - 3.4.2 AirSAR L-Band data pre-processing - Object-Oriented methodology - Refined Lee filtering - Segmentation - Object-based information - Shape-based features - Texture-based feature - 3.4.3 AirSAR L-Band data classification by TexSOM - 3.4.4 PolSOM and TexSOM AirSAR classification: results comparison - 3.4.5 RADARSAT-2 data pre-processing and classification - 3.4.6 PolSOM and TexSOM RADARSAT-2 classification: results comparison - 4 Conclusions - A AirSAR data compression/decompression equations - Bibliographyen
format.extent7074994 bytes-
relation.isbasedonJ. Lee, M. Grunes, T. Ainsworth, L. Du, D. Schuler, and S. Cloude,”Unsupervised classification using polarimetric decomposition and the complex wishart classifier,en
relation.isbasedonS. Cloude and E. Pottier, “A review of target decomposition theorems in radar polarimetry,en
subjectself organizing mapen
titlePolSOM and TexSOM in polarimetric SAR classificationen
typeDoctoral thesisen
degree.disciplineFacoltà di ingegneriaen
degree.grantorUniversità degli studi di Roma Tor Vergataen
relation.referenceW. Boerner, A.-Q. Yan, and Y. Yamaguchi, “On the basic principles of radar polarimetry: the target characteristic polarization state theory of Kennaugh, Huynens polarization fork concept, and its extension to the partially polarized case,en
relation.referenceJ. Van Zyl, “Unsupervised classification of scattering behavior using radar polarimetry data,en
relation.referenceS. Chandrasekhar, Radiative Transfer. Dover, 1960.en
relation.referenceJ. Huynen, Phenomenological Theory of Radar Targets. PhD thesis,en
relation.referenceW. L. Cameron and L. K. Leung, “Feature motivated polarization scattering matrix decomposition,en
date.dateofdefenseA.A. 2008/2009en
Appears in Collections:Tesi di dottorato in ingegneria

Files in This Item:

File Description SizeFormat
C_Putignano_PhD_Thesis.pdf6909KbAdobe PDFView/Open

Show simple item record

All items in DSpace are protected by copyright, with all rights reserved.