Deep learning algorithm for urban feature extraction using SAR data (Soft copy is also available)
Pithva, Nikunj M. (PG180613)
Deep learning algorithm for urban feature extraction using SAR data (Soft copy is also available) - 2020 - xvii,51p.
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 xiii
List of tables xvii
Chapter 1 Introduction 1
1.1 Background and Significance 1
1.1.1 Burning Issues in Urban Studies 1
1.1.2 SAR Urban Application and Benefits2
1.1.3 Polarization 2
1.1.4 Circular Polarization (Hybrid Polarimetry) 3
1.1.5 Geometric Properties 4
1.1.6 Speckle Suppression Filters 5
1.1.7 Backscatter and its coefficient6
1.2 Aim and Objective7
Chapter 2 Literature Review 8
2.1 Different Products used in Literature survey 8
Chapter 3 Methodology 12
3.1 Four phase in methodology12
3.1.1 Phase 1 13
3.1.2 Phase 2 14
3.1.3 Phase 3 15
3.1.4 Phase 4 16
3.2 Study Area and Data Description17
Chapter 4 Analysis and Results 18
4.1 Data Pre-processing18
4.2 Back Scatter co-efficient 23
4.2.1 Mean back scatter co-efficient 24
4.2.2 Variation of Sigma naught for Different Features 25
4.2.3 Variation of Sigma naught for Urban Features 26
4.3 Analysis on Processed Image 27
4.4 Image Fusion 29
4.4.1 Qualitative Test29
4.4.2 Quantitative Statistical Test32
4.5 Segmentation 33
4.5.1 Image segmentation: Multi resolution Segmentation 34
4.6 Deep Learning Architecture 36
4.6.1 Diagram of Convolution neural Network36
4.6.2 Result of CNN for Fused Image (256 X 256) 38
4.6.3 Result of CNN for Fused Image (350 X 350) 39
4.6.4 Result of CNN for amplitude Image (256 X 256) 40
4.6.5 Result of CNN for amplitude Image (350 X 350) 42
4.6.6 Extracting Urban features from the Image 44
5 Conclusion and Future Scope46
Appendix 1: Convolution Neural Network code 49
Plagiarism Report Copy 51
MG TH-0197 / PIT
Deep learning algorithm for urban feature extraction using SAR data (Soft copy is also available) - 2020 - xvii,51p.
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 xiii
List of tables xvii
Chapter 1 Introduction 1
1.1 Background and Significance 1
1.1.1 Burning Issues in Urban Studies 1
1.1.2 SAR Urban Application and Benefits2
1.1.3 Polarization 2
1.1.4 Circular Polarization (Hybrid Polarimetry) 3
1.1.5 Geometric Properties 4
1.1.6 Speckle Suppression Filters 5
1.1.7 Backscatter and its coefficient6
1.2 Aim and Objective7
Chapter 2 Literature Review 8
2.1 Different Products used in Literature survey 8
Chapter 3 Methodology 12
3.1 Four phase in methodology12
3.1.1 Phase 1 13
3.1.2 Phase 2 14
3.1.3 Phase 3 15
3.1.4 Phase 4 16
3.2 Study Area and Data Description17
Chapter 4 Analysis and Results 18
4.1 Data Pre-processing18
4.2 Back Scatter co-efficient 23
4.2.1 Mean back scatter co-efficient 24
4.2.2 Variation of Sigma naught for Different Features 25
4.2.3 Variation of Sigma naught for Urban Features 26
4.3 Analysis on Processed Image 27
4.4 Image Fusion 29
4.4.1 Qualitative Test29
4.4.2 Quantitative Statistical Test32
4.5 Segmentation 33
4.5.1 Image segmentation: Multi resolution Segmentation 34
4.6 Deep Learning Architecture 36
4.6.1 Diagram of Convolution neural Network36
4.6.2 Result of CNN for Fused Image (256 X 256) 38
4.6.3 Result of CNN for Fused Image (350 X 350) 39
4.6.4 Result of CNN for amplitude Image (256 X 256) 40
4.6.5 Result of CNN for amplitude Image (350 X 350) 42
4.6.6 Extracting Urban features from the Image 44
5 Conclusion and Future Scope46
Appendix 1: Convolution Neural Network code 49
Plagiarism Report Copy 51
MG TH-0197 / PIT