跳到主要內容

臺灣博碩士論文加值系統

(34.204.180.223) 您好!臺灣時間:2021/07/31 18:15
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:張懿聖
研究生(外文):Yi-Sheng Chang
論文名稱:街景清道夫:自Google街景圖中偵測與去除車輛
論文名稱(外文):Street Sweeper: Detection and Removal of Vehicles in Google Street View Images
指導教授:朱威達
指導教授(外文):Wei-Ta Chu
口試委員:郭景明施皇嘉王昱舜朱威達
口試委員(外文):Jing-Ming GuoHuang-Chia ShihYu-Shuen WangWei-Ta Chu
口試日期:2012-07-09
學位類別:碩士
校院名稱:國立中正大學
系所名稱:資訊工程研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2012
畢業學年度:100
語文別:英文
論文頁數:56
中文關鍵詞:GrabcCut種子找尋方向性補繪
外文關鍵詞:GrabcCutseed findingdirectional inpainting
相關次數:
  • 被引用被引用:0
  • 點閱點閱:362
  • 評分評分:
  • 下載下載:68
  • 收藏至我的研究室書目清單書目收藏:0
隨著地圖服務越來越普遍,地圖服務的業者在此服務上提供的資料也越來越詳細以迎合使用者的需求。新型態的地圖服務–街景圖–提供了街道層級的視覺資訊,然而這也同時帶來了隱私方面的疑慮;為了保護用路人的隱私,我們提出了一個能自動偵測與移除汽機車的系統。即便地圖服務的業者目前提供把街景圖中車牌與人臉模糊化的方案,我們認為殘留於街景圖上的特徵,如印在車身上的車牌號碼、電話號碼,甚至是機車騎士的身型,仍然會洩漏用路人的隱私。在輸入一連串由街景車沿街拍攝的街景圖後,我們的系統將先選出感興趣區域(region of interest)以避免遭受週遭景物,如建築、行道樹,的影響。在進行圖像間的動量分析(motion analysis)後,我們的系統將會選出可能為前景或背景的種子以輸入給GrabCut圖像切割系統。這項設計避免了GrabCut系統預設的人工輸入。在移除車輛後,我們套用一個基於樣本的影像補繪(exemplar-based inpainting)方法來填補缺失區域,其填補的優先權計算被特別設計為要考慮紋理方向以填補出合適的結果。
在實驗中,我們分析了實驗組的特性,評量了我們的車輛偵測程序,並且將我們的補繪方法與其他補繪方法作比較。實驗結果證明我們的方法的確能更加保護用路人的隱私,而且我們採用的特性確實對車輛偵測與補繪有所幫助。
As the map service becomes commonly used, map service providers provide more detailed information on map to satisfy the users’ needs. Street view service, as a brand new kind of map service, provides street level visual information. However, this service also causes the problem of privacy leakage. To protect individual’s and company’s privacy, we present a system that automatically detects vehicles and riders associated with their motorbikes/bicycles, and removes them as if they had never been there. Although street view service providers have made efforts on blurring human faces and license plates, we argue that the remaining features, such as license numbers and phone numbers printed on car bodies, and shapes of riders, could still leak privacy. Given a sequence of street view images that were consecutively captured by the camera car driving along a way, the developed system first determines the region of interest by filtering out noisy objects such as building and trees on the roadside. By conducting motion analysis between images, this system then determines candidate foreground seeds and background seeds, which are later fed to a GrabCut image segmentation module. This design avoids human input that is originally demanded in the conventional Grabcut approach. After removing detected vehicles, an exemplar-based inpainting method, with special designs on determination of filling priority and direction of texture propagation, is adopted to make pleasing reconstruction results.
In the experiments, we analyze properties of our datasets, evaluate performance of the vehicle detection process, and compare our inpainting method with others. The experimental results show that our system protects drivers’ privacy than the methods currently used by map service providers.
誌謝 i
摘要 ii
Abstract iii
LIST OF FIGURES vi
LIST OF TABLES viii
LIST OF ALGORITHMS ix
Chapter 1 INTRODUCTION 1
1.1 Motivation 1
1.2 Challenges 2
1.3 Contributions 3
1.4 Thesis Organization 4
Chapter 2 RELATED WORKS 5
2.1 Applications on Street View Images 5
2.2 Inpainting 7
2.3 Summary 8
Chapter 3 Vehicle Detection 9
3.1 Framework 9
3.2 Road Detection 10
3.3 Foreground Extraction 13
3.3.1 Introduction of GrabCut 13
3.3.2 Foreground Seed Selection 14
3.3.3 Background Seed Selection 16
3.3.4 Result of Foreground Extraction 18
3.4 Morphological Operations 19
3.5 Summary 21
Chapter 4 Removal of Vehicles 23
4.1 Determining Filling Priority 23
4.2 Texture Propagation 26
4.3 Summary 28
Chapter 5 Experimental Results 30
5.1 Datasets 30
5.2 Performance of Vehicle Detection 33
5.3 Results of Vehicle Removal 35
5.4 Performance of Inpainting 36
5.5 User Study 38
5.6 Discussion 39
5.7 Summary 40
Chapter 6 Conclusion and Future Work 41
6.1 Conclusion 41
6.2 Future Work 41
REFERENCES 43
APPENDIX 45
[1] A. Frome, G. Cheung, A. Abdulkader, M. Zennaro, B. Wu, A. Bissacco, H. Adam, H. Neven, and L. Vincent. “Large-scale Privacy Protection in Google Street View,” Proceedings of IEEE International Conference on Computer Vision, pp. 2373 – 2380, 2009
[2] L. Vincent. “Taking Online Maps Down to Street Level,” Computer, 40, 12 (December), pp. 118–120, 2007.
[3] Y. Yoshimoto, T. H. Dang, A. Kimura, F. Shibata, and H. Tamura. “Interaction Design of 2d/3d Map Navigation on Wall and Tabletop Displays,” Proceedings of ACM International Conference on Interactive Tabletops and Surfaces, pp. 254–255, 2011.
[4] R. Guy and K. Truong. “Crossingguard: Exploring Information Content in Navigation Aids for Visually Impaired Pedestrians,” Proceedings of ACM annual conference on Human Factors in Computing Systems, pp. 405–414, 2012.
[5] J. Kopf, B. Chen, R. Szeliski, and M. Cohen, “Street Slide: Browsing Street Level Imagery,” ACM Transactions on Graphics 29, 4 (July), Article 96, 2010.
[6] A. Flores and S. Belongie, “Removing Pedestrians from Google Street View Images,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 53–58, 2010.
[7] B. Leibe, A. Leonardis, and B. Schiele. “Robust Object Detection with Interleaved Categorization and Segmentation,” International Journal of Computer Vision, 77, pp. 259–289, 2008.
[8] D. Tschumperle and R. Deriche. “Vector-valued Image Regularization with PDEs: A Common Framework for Different Applications,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 27, 4 (April), pp. 506–517, 2005.
[9] A. Criminisi, P. Perez, and K. Toyama. “Region Filling and Object Removal by Exemplar-based Image Inpainting,” IEEE Transactions on Image Processing, 13, 9 (September), pp. 1200–1212, 2004.
[10] O. Le Meur, J. Gautier, and C. Guillemot. “Examplar-based Inpainting Based on Local Geometry,” Proceedings of IEEE International Conference on Image Processing, pp. 3401–3404, 2011.
[11] C. Rother, V. Kolmogorov, and A. Blake. “Grabcut: Interactive Foreground Extraction Using Iterated Graph Cuts,” ACM Transactions on Graphics, 23, 3 (August), pp. 309–314, 2004.
[12] Y. Y. Boykov and M.-P. Jolly. “Interactive Graph Cuts for Optimal Boundary & Region Segmentation of Objects in N-d Images,” Proceedings of International Conference on Computer Vision, pp. 105–112, 2001.
[13] S. Di Zenzo. “A Note on the Gradient of a Multi-image.” Computer Vision, Graphics, and Image Processing, 33, 1 (January), pp. 116–125, 1986
[14] Y. Wexler, E. Shechtman, and E. Irani. ”Space-time Completion of Video,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 29, 3(March), pp. 463–476, 2007
[15] N. Komodakis and G. Tziritas. “Image Completion using Efficient Belief Propagation via Priority Scheduling and Dynamic Pruning,” IEEE Transactions on Image Processing 16, 11 (November), pp. 2649–2661, 2007.
[16] A. Telea. “An Image Inpainting Techniques based on the Fast Matching Method,” Journal of Graphics Tools 9, 1 (January), pp. 25–36, 2004.
[17] M. M. Chen, G. X. Zhang, N. J. Mitra, X. Huang, and S.M. Hu. “Global Contrast based Salient Region Detection,” Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pp. 409-416, 2011.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top