| ABSTRACT | 第4页 |
| 摘要 | 第5-7页 |
| ACKNOWLEDGEMENTS | 第7-10页 |
| Ⅰ. Introduction and Research Background | 第10-18页 |
| Ⅰ.1. REASONS AND SIGNIFICANCE OF DISSERTATION TOPIC: | 第11-14页 |
| Ⅰ.1.1. Problem Statement | 第11-13页 |
| Ⅰ.1.2. Motivation | 第13-14页 |
| Ⅰ.2. LITERATURE REVIEW: | 第14-16页 |
| Ⅰ.2.1. Urban 3D modeling trend | 第14-15页 |
| Ⅰ.2.2. Deep learning and object detection tendency | 第15页 |
| Ⅰ.2.3. Object removal development | 第15-16页 |
| Ⅰ.3. RESEARCH CONTENT: | 第16-18页 |
| Ⅰ.3.1. Research objectives | 第16页 |
| Ⅰ.3.2. Outline of Thesis | 第16-18页 |
| Ⅱ. Three dimensional city modeling | 第18-29页 |
| Ⅱ.1. INTRODUCTION TO URBAN 3D MODELS: | 第19-23页 |
| Ⅱ.1.1. Definitions of 3D city models | 第19-20页 |
| Ⅱ.1.2. Categorization of 3D city models | 第20-21页 |
| Ⅱ.1.3. Comparison of 3D city modeling techniques | 第21-23页 |
| Ⅱ.2. CREATION OF 3D CITY MODELS: | 第23-25页 |
| Ⅱ.2.1. Data collection | 第23页 |
| Ⅱ.2.2. Data processing | 第23-24页 |
| Ⅱ.2.3. Modeling workflow | 第24-25页 |
| Ⅱ.3. APPLICATIONS OF 3D CITY MODELS: | 第25-27页 |
| Ⅱ.3.1. Safety and security applications | 第25-26页 |
| Ⅱ.3.2. Economic and leisure application | 第26-27页 |
| Ⅱ.3.3. Research and development application | 第27页 |
| Ⅱ.4. LIMITATIONS OF 3D CITY MODELS: | 第27-29页 |
| Ⅲ. Car Detection using Deep Learning | 第29-46页 |
| Ⅲ.1. IMAGE CLASSIFICATION WITHCNNs: | 第30-37页 |
| Ⅲ.1.1. Genesis of CNNs and Deep Learning | 第31-32页 |
| Ⅲ.1.2. Sparse coding for image classification | 第32-35页 |
| Ⅲ.1.3. Operation Principles of Deep Learning in image classification | 第35-37页 |
| Ⅲ.2. CHOICE OF THE CLASSIFICATION APPROACH AND THE FEATURES SPACE: | 第37-40页 |
| Ⅲ.2.1. Statistical Learning Framework | 第37-38页 |
| Ⅲ.2.2. Feature Spaces Determination | 第38-39页 |
| Ⅲ.2.3. Choice of the object detection Framework | 第39-40页 |
| Ⅲ.3. DESIGN OF THE DEEP LEARNING CAR DETECTOR: | 第40-46页 |
| Ⅲ.3.1. Car Detection System Architecture | 第40-42页 |
| Ⅲ.3.2. Deep Network Training | 第42-44页 |
| Ⅲ.3.3. Deep Learning Car Detection Process | 第44-46页 |
| Ⅳ. Car Removal by Exemplar Based Inpainting | 第46-57页 |
| Ⅳ.1. OBJECT REMOVAL CONTEXT: | 第47-51页 |
| Ⅳ.1.1. Introduction to object removal | 第47-48页 |
| Ⅳ.1.2. Image Inpainting challenges and applications | 第48-49页 |
| Ⅳ.1.3. Object removal techniques | 第49-51页 |
| Ⅳ.2. IMPLEMENTATION OF EXEMPLAR BASED INPAINTING CAR REMOVAL SYSTEM: 49 | 第51-57页 |
| Ⅳ.2.1. Computation of the filling priority and pertinence of patches | 第51-53页 |
| Ⅳ.2.2. Texture and structure information propagation | 第53-54页 |
| Ⅳ.2.3. Car removal complete process | 第54-57页 |
| Ⅴ. Experimental Results and Discussion | 第57-69页 |
| Ⅴ.1. RESULTS& EVALUATION OF THE CAR DETECTION SYSTEM: | 第58-62页 |
| Ⅴ.1.1. Car detection system experimental results | 第58-61页 |
| Ⅴ.1.2. Car detection results discussion | 第61-62页 |
| Ⅴ.2. RESULTS &EVALUATION OF THE CAR REMOVAL SYSTEM: | 第62-65页 |
| Ⅴ.2.1. Car removal system experimental results | 第62-64页 |
| Ⅴ.2.2. Car removal results discussion | 第64-65页 |
| Ⅴ.3. THREE DIMENSIONAL MODELING VALIDATION OF THE PROPOSED APPROACH: | 第65-69页 |
| Ⅵ. Conclusion | 第69-76页 |
| Ⅵ.1. SUMMARY OF THE RESEARCH AND OBJECTIVES COMPLETION: | 第70-72页 |
| Ⅵ.2. CONTRIBUTIONS AND ASSESSMENT OF THE RESEARCH | 第72-74页 |
| Ⅵ.3. PERSPECTIVE WORKS | 第74-76页 |
| Ⅶ. Bibliography | 第76-78页 |
