Feature extraction from point cloud generated using UAV based photogrammetry (Softcopy is also available)
Soni, Siddharth (PT201517)
Feature extraction from point cloud generated using UAV based photogrammetry (Softcopy is also available) - 2019 - ii,iv,xv,60p.
Table of Contents
Certificate of Original Authorship i
Certificate of Dissertation - Advisor’s consent iii
Certificate of Dissertation - Program Chair v
Acknowledgments vii
Abstract ix
List of figures xi
List of tables xv
1 INTRODUCTION 1
1.1 Drone Survey 2
1.1.1 Orthomosaic 3
1.1.2 Point Cloud 3
1.1.3 DEM and DSM 3
1.1.4 Textured Mesh 3
1.2 Research Problems 4
1.2.1 Complex Scene 4
1.2.2 Lack of Conformity 4
1.2.3 Complex Geometry 4
1.3 Goal and Objectives 5
1.4 Scope and Limitations 5
1.5 Need for Study 6
1.6 Thesis Outline 6
2 LITERATURE REVIEW 9
2.1 Building Detection from Satellite Images 9
2.2 Building Detection from Aerial Images 10
2.3 Building Detection from High Density Point Clouds 11
2.4 Conclusions 12
3 DATA ACQUISITION AND PROCESSING IN DRONE SURVEY 13
3.1 Ground Control Points 13
3.2 Flight Parameters 14
3.2.1 Camera Specifications 14
3.2.2 Flight Path 15
3.2.3 Flight Height 16
3.2.4 Front and Side Lap 16
3.3 Understanding Spatial Resolution with Drones 17
3.4 Processing of Data 18
3.4.1 Preprocessing of Images 19
3.4.2 Structure from Motion Algorithm 19
3.4.3 3D Scene Reconstruction 21
4 POINT CLOUD PROCESSING 23
4.1 Types of Point Cloud 23
4.1.1 LiDAR 24
4.1.2 Digital Aerial Photogrammetry 25
4.2 Point Feature Representation 26
4.2.1 Normal Estimation 26
4.2.2 Curvature 27
4.2.3 Point Feature Histograms 28
4.3 Point Cloud Segmentation 28
4.4 Plane Extraction in Point Clouds 28
4.4.1 Region Growing based Method 29
4.4.2 Random Sample Consensus 29
4.4.3 Hough Transform 30
4.5 Multi-view Registration of Point Cloud 30
5 RESEARCH METHODOLOGY 31
5.1 Forming Neighbourhoods 31
5.2 Tabulate Local Feature Descriptors 32
5.2.1 Surface Normals 32
5.2.2 Surface Density 32
5.2.3 Volume Density 33
5.2.4 Curvature 33
5.2.5 Roughness 33
5.3 Segregate Ground Points and Non Ground Points 33
5.3.1 Cloth Simulation Filter 33
5.3.2 Progressive Morphological Filter 34
5.4 Plane Segmentation using RANSAC 35
5.5 Identifying Connected Components 35
5.6 Edge Points Detection 36
5.7 Boundary Approximation Using Regularization Constraints 36
5.8 Roof Modelling 36
5.9 Generating DEM and Contour from Ground Points 37
6 DATASET 39
6.1 Initial Image Positions 40
6.2 Geolocation Details 41
6.3 Point Cloud Densification and DSM Details 41
6.4 Subset of Data 42
7 RESULTS AND OUTPUTS 43
7.1 Point Cloud Data Formats 43
7.2 Local Feature Representations 44
7.2.1 Surface Normals 44
7.2.2 Number of Neighbours 45
7.2.3 Curvature 46
7.2.4 Roughness 47
7.3 Segregating Ground Points and Non-Ground Points 47
7.4 Plane Extraction 48
7.5 Edge Point Detection and Boundary Regularization 49
7.6 DEM and Contour Generation 50
7.7 Reconstructed Urban Scene 52
7.8 Comparison with other methods 53
8 CONCLUSION 55
References 57
Appendix 59
MG TH-0183 / SON
Feature extraction from point cloud generated using UAV based photogrammetry (Softcopy is also available) - 2019 - ii,iv,xv,60p.
Table of Contents
Certificate of Original Authorship i
Certificate of Dissertation - Advisor’s consent iii
Certificate of Dissertation - Program Chair v
Acknowledgments vii
Abstract ix
List of figures xi
List of tables xv
1 INTRODUCTION 1
1.1 Drone Survey 2
1.1.1 Orthomosaic 3
1.1.2 Point Cloud 3
1.1.3 DEM and DSM 3
1.1.4 Textured Mesh 3
1.2 Research Problems 4
1.2.1 Complex Scene 4
1.2.2 Lack of Conformity 4
1.2.3 Complex Geometry 4
1.3 Goal and Objectives 5
1.4 Scope and Limitations 5
1.5 Need for Study 6
1.6 Thesis Outline 6
2 LITERATURE REVIEW 9
2.1 Building Detection from Satellite Images 9
2.2 Building Detection from Aerial Images 10
2.3 Building Detection from High Density Point Clouds 11
2.4 Conclusions 12
3 DATA ACQUISITION AND PROCESSING IN DRONE SURVEY 13
3.1 Ground Control Points 13
3.2 Flight Parameters 14
3.2.1 Camera Specifications 14
3.2.2 Flight Path 15
3.2.3 Flight Height 16
3.2.4 Front and Side Lap 16
3.3 Understanding Spatial Resolution with Drones 17
3.4 Processing of Data 18
3.4.1 Preprocessing of Images 19
3.4.2 Structure from Motion Algorithm 19
3.4.3 3D Scene Reconstruction 21
4 POINT CLOUD PROCESSING 23
4.1 Types of Point Cloud 23
4.1.1 LiDAR 24
4.1.2 Digital Aerial Photogrammetry 25
4.2 Point Feature Representation 26
4.2.1 Normal Estimation 26
4.2.2 Curvature 27
4.2.3 Point Feature Histograms 28
4.3 Point Cloud Segmentation 28
4.4 Plane Extraction in Point Clouds 28
4.4.1 Region Growing based Method 29
4.4.2 Random Sample Consensus 29
4.4.3 Hough Transform 30
4.5 Multi-view Registration of Point Cloud 30
5 RESEARCH METHODOLOGY 31
5.1 Forming Neighbourhoods 31
5.2 Tabulate Local Feature Descriptors 32
5.2.1 Surface Normals 32
5.2.2 Surface Density 32
5.2.3 Volume Density 33
5.2.4 Curvature 33
5.2.5 Roughness 33
5.3 Segregate Ground Points and Non Ground Points 33
5.3.1 Cloth Simulation Filter 33
5.3.2 Progressive Morphological Filter 34
5.4 Plane Segmentation using RANSAC 35
5.5 Identifying Connected Components 35
5.6 Edge Points Detection 36
5.7 Boundary Approximation Using Regularization Constraints 36
5.8 Roof Modelling 36
5.9 Generating DEM and Contour from Ground Points 37
6 DATASET 39
6.1 Initial Image Positions 40
6.2 Geolocation Details 41
6.3 Point Cloud Densification and DSM Details 41
6.4 Subset of Data 42
7 RESULTS AND OUTPUTS 43
7.1 Point Cloud Data Formats 43
7.2 Local Feature Representations 44
7.2.1 Surface Normals 44
7.2.2 Number of Neighbours 45
7.2.3 Curvature 46
7.2.4 Roughness 47
7.3 Segregating Ground Points and Non-Ground Points 47
7.4 Plane Extraction 48
7.5 Edge Point Detection and Boundary Regularization 49
7.6 DEM and Contour Generation 50
7.7 Reconstructed Urban Scene 52
7.8 Comparison with other methods 53
8 CONCLUSION 55
References 57
Appendix 59
MG TH-0183 / SON