 | Avbelj J. Fusion of hyperspectral images and digital surface models for urban object extraction: Diss. … Dr.-Ing. / Deutsches Zentrum für Luft- und Raumfahrt, Institut für Methodik der Fernerkundung, Oberpfaffenhofen - Köln: DLR, 2016. - 135 p.: ill., tab. - (Forschungsbericht; 2016-07). - Res. also Germ. - Bibliogr.: p.129-133. - ISSN 1434-8454
Шифр: (Pr 1120/2016-07) 02
|
1 Introduction ............................................... 9
1.1 Scientific Relevance of the Topic .......................... 9
1.2 Objectives and Focus of the Thesis ........................ 10
1.3 Outline ................................................... 11
2 Theoretical Background .................................... 13
2.1 HyperspectraJ Imaging ..................................... 13
2.1.1 Terminology and Basic Principles of HSI ............ 13
2.1.2 Distortions in HSI and Their Characteristics ....... 15
2.2 Digital Elevation Models .................................. 19
2.2.1 Terminology ........................................ 19
2.2.2 DEM Generation and Their Characteristics ........... 19
2.3 Image Fusion in Urban Areas ............................... 22
2.3.1 Availability of the Multi-View and Hyperspectral
Images ............................................. 23
2.3.2 Potential and Challenges of the HSI and DSM
Fusion ............................................. 23
3 Advances in HSI and DSM Fusion for Building Extraction .... 25
3.1 Building Extraction from Remote Sensing Data .............. 25
3.1.1 General Workflow of Geometrical Building
Extraction ......................................... 26
3.1.2 Least Squares Adjustment of Rectilinear Building
Outlines ........................................... 28
3.2 Remote Sensing and Scale Space ............................ 29
3.3 Evaluation of Building Outline Extraction ................. 31
3.3.1 Wild West in Evaluation Techniques for Extracted
Building Polygons .................................. 31
3.3.2 Overview of Indices ................................ 32
3.3.3 What is Ground Truth? .............................. 33
3.4 Summary of Research Voids ................................. 34
4 Fusion of HSI and DSM ..................................... 35
4.1 Edge in an Image .......................................... 35
4.1.1 Edge Detection and Edge Probability in an Image .... 38
4.1.2 Scale Space Representation ......................... 41
4.1.3 Implementation of the Discrete Linear Scale Space
and Scale Space Derivatives ........................ 42
4.2 Bayesian Fusion of HSI and DSM Based on Edges ............. 44
4.2.1 Abundance and Probability of an Edge ............... 41
4.2.2 DSM Derived Probability of an Edge ................. 45
4.2.3 Gaussian Mixture Model ............................. 45
4.3 Information Content Relation of HSI and DSM ............... 46
4.3.1 Weighting by Prior Knowledge ....................... 46
4.3.2 Weighting by Data-Based Confidence Level ........... 46
5 Building Outline Extraction and Adjustment ................ 47
5.1 Approximate Building Outline Determination ................ 47
5.1.1 Building Region Extraction ......................... 47
5.1.2 Building Polygon Creation and Selection ............ 49
5.2 Joined use of HSI and DSM for Building Polygon
Estimation ................................................ 51
5.2.1 Mathematical Model for Rectilinear Polygon ......... 52
5.2.2 "Weights in the Adjustment ......................... 56
5.3 A New Metric for Evaluation of Polygons and Line
Segments (PoLiS) .......................................... 58
5.3.1 Measures for Quantification of Similarity .......... 59
5.3.2 Definition of the PoLiS Metric ..................... 66
5.3.3 Characteristics of the PoLiS Metric ................ 67
6 Case Studies .............................................. 75
6.1 Data Description and Preprocessing ........................ 75
6.1.1 HSI Sensors and Images ............................. 75
6.1.2 DSM from Stereo Images and LiDAR Point Clouds ...... 78
6.1.3 Implementation and Setting of Parameters for
Proposed Methods ................................... 78
6.2 Experiments on Scale Space for Edge Detection ............. 79
6.2.1 Test Dataset ....................................... 79
6.2.2 Parameter Settings ................................. 80
6.2.3 Results and Discussion ............................. 80
6.2.4 Summary of Experiments on Scale Space for Edge
Detection .......................................... 85
6.3 Experiment on Building Polygon Selection and Adjustment ... 85
6.3.1 Test Dataset ....................................... 86
6.3.2 Parameter Settings ................................. 87
6.3.3 Results and Discussion of the Model Selection
Experiment ......................................... 88
6.3.4 Results and Discussion of the Adjustment ........... 92
6.3.5 -Summary of Experiment on BP Selection and
Adjustment ......................................... 96
6.4 Experiments on RS Images .................................. 97
6.4.1 Test Dataset ....................................... 98
6.4.2 Parameter Settings and Preprocessing ............... 99
6.4.3 Results and Discussion of the Experiment on the
Small Area ........................................ 100
6.4.4 Results and Discussion of the Experiment on the
Large Area ........................................ 109
6.4.5 Summary of Experiments on RS Images ............... 115
7 Conclusions and Outlook .................................. 119
7.1 Conclusions .............................................. 119
7.2 Outlook .................................................. 120
Acronyms ................................................. 123
List of Figures .......................................... 125
List of Tables ........................................... 127
Bibliography .................................................. 129
|
|
