跳到主要內容

臺灣博碩士論文加值系統

(44.210.83.132) 您好!臺灣時間:2024/05/29 14:22
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

我願授權國圖
: 
twitterline
研究生:邱子軒
研究生(外文):Tzu-Hsuan Chiu
論文名稱:主從式自動追蹤視訊監控系統
論文名稱(外文):A Master-Slave Auto-trackig Visual Surveillance System
指導教授:洪一平洪一平引用關係
學位類別:碩士
校院名稱:國立臺灣大學
系所名稱:資訊工程學研究所
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:2009
畢業學年度:97
語文別:英文
論文頁數:54
中文關鍵詞:監控視訊主從式攝影機控制
外文關鍵詞:surveillancemaster-slavecamera control
相關次數:
  • 被引用被引用:1
  • 點閱點閱:237
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
在這篇論文裡,我們提出一套主從式智慧型監控系統。此系統硬體由兩支攝影機組成,其中一台為固定式場域監控攝影機,一台為高速球型攝影機。此系統可以監控廣泛區域,並同時取得監控場景中物體的高清晰影像,以作為進階影像分析的基礎,例如:人臉偵測,與車牌辨識,此外還可以用高速球型攝影機對監控場景中的物體進行自動追蹤。為了達到上述功能,我們需要達到以下的要求:1.高速球型攝影機必需能夠精確轉向固定式場域監控攝影機監看畫面中的任意位置。2.在固定式場域監控攝影機畫面中能追蹤物體,並控制高速球型攝影機進行旋轉追蹤。本系統的主要貢獻在於我們幾乎將校正流程幾乎全自動化,並且我們在追蹤的過程中有將高速球形攝影機旋轉的時間列入考慮。此系統在第1項要求達到的精準度誤差隨固定式場域監控攝影機與高速球型攝影機的擺放方式不同(共鏡心或非共鏡心)與倍數不同而介於3到45個畫素之間。在第2項要求達到的高速球型攝影機追蹤誤差則隨高速球型攝影機的倍數不同,以及追蹤物體的移動速度不同而介於30到75個像素之間。本論文並對人手動控制高速球型攝影機追蹤物體及系統自動控制的誤差作了比較,結論是系統的控制比人手動控制精準許多。
In this thesis, we proposed a master-slave auto-tracking surveillance system. This system is consisted of two cameras, one of them is wide angle fixed camera, and another one is speed dome camera. This system is able to monitor a wide area and gets high quality images of objects in the monitored scene for further analysis, such as face recognition and vehicle plate recognition. Besides, the speed dome of this system is capable of automatically tracking objects in the monitored scene. In order to achieve the functions mentioned above, we have to satisfy the following demands. First, the speed dome camera must be able to turn to any place in the image of wide angle fixed camera. Second, tracking objects in the image sequences of wide angle fixed camera, and then controls the speed dome camera to track them. The main contributions of this work are that we make the calibration process for the first demand almost automatic, and we consider the turning time of speed dome camera when tracking objects. The error of this system in the first demand is between 3 to 45 pixels depending on how the wide angle fixed camera and speed dome camera are placed (concentric or non-concentric) and the zoom factor. The error of this system in the second demand is between 30 to 75 pixels, and it is depending on zoom factor of speed dome camera, and the velocity of the tracked objects. In this paper, we have also compared the error between human manually tracking and system auto-tracking. The conclusion is that system auto-tracking is much better.
Abstract ix
List of Figures xiii
List of Tables xv
1 Introduction 1
2 Background 5
2.1 Pinhole Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Homography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Concentric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.2 Coplaner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.1 Under Lying Motion Model . . . . . . . . . . . . . . . . . . . . 10
2.3.2 Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.3 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Related Works 17
3.1 Hybrid sensors calibration : Application to pattern recognition and tracking 17
3.2 Real-Time Orientation of a PTZ-Camera Based on Pedestrian . . . . . . . 19
4 Camera Calibration 23
4.1 Calibration between Wide Angle Fixed Camera and Speed Dome Camera 23
4.1.1 Concentric Case . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.1.2 Non-Concentric Case . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Speed Dome Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.2.1 Ideal Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.2.2 Look-up Table . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.2.3 Conclusion on Camera Calibration . . . . . . . . . . . . . . . . . 35
5 Object Tracking 39
5.1 Track Specified Object in the Image Sequences of Wide Angle Fixed
Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.1.1 Background Subtraction . . . . . . . . . . . . . . . . . . . . . . 42
5.1.2 Tracking and Prediction using Kalman Filter . . . . . . . . . . . 43
5.2 Control the Speed Dome toward the Object . . . . . . . . . . . . . . . . 44
6 Experiments 47
6.1 Camera Calibration Experiment . . . . . . . . . . . . . . . . . . . . . . 47
6.1.1 Concentric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.1.2 Non-concentric . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
6.2 Tracking Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7 Conclusions and Future Works 51
7.1 Conclsions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
7.2 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Bibliography 53
[1] J. Badri, C. Tilmant, J.-M. Lavest, Q.-C. Pham, and P. Sayd. Hybrid sensors calibration : Application to pattern recognition and tracking. WISP, 2007. 18
[2] C.-S. Chen, C.-K. Yu, and Y.-P. Hung. New calibration-free approach for augmented reality based on parameterized cuboid structure. ICCV, 1999. 25, 27, 29
[3] J. Foote, Q. Liu, D. Kimber, P. Chiu, , and F. Zhao. Reach-through-the-screen: A new metaphor for remote collaboration. Advances in Multimedia Information Processing, 2004. 21
[4] T. Hoedl, D. Brandt, and U. S. znd M. Wiggenhagen. Real-time orientation of a ptz-camera based on pedestrian. ISPRS, 2008. 19
[5] R. Horaud, D. Knossow, and M. Michaelis. Camera cooperation for acheiving visual attention. Machine Vision and Application, 2006. 20
[6] R. Kalman. A new approach to linear ltering and prediction problems. Journal of Basic Engineering, 1960. 10
[7] R. Kalman and R. Bucy. New results in linear ltering and prediction theory. American Society of Mechanical Engineers, 1961. 10
[8] N. K. Kanhere, S. T. Bircheld, and W. A. Sarasua. Vehicle segmentation and tracking in the presence of occlusions. TRB, 2006. 26
[9] K. Kim, T. H. halidabhongse, D. Harwood, and L. S. Davis. Background modeling and subtraction by codebook construction. ICIP, 2004. 42
[10] K. Kim, T. H. Chalidabhongse, D. Harwood, and L. Davis. Real-time foreground–background segmentation using codebook model. Real-Time Imaging, 2005. 42
[11] N. Krahnstoever, T. Yu, S.-N. Lim, K. Patwardhan, and P. Tu. Collaborative real-time control of active cameras in collaborative real-time control of active cameras in large scale surveillance systems. M2SFA2, 2008. 20
[12] M. L., M. L., and R. C. Dual camera system for face detection in unconstrained environments. ICIP, 2003. 21
[13] C. Liao, Q. Liu, D. Kimber, P. Chiu, J. Foote, and L. W. and. Shared interactive video for teleconferencing. 21
[14] Q. Liu, D. Kimber, J. Foote, L. Wilcox, and J. Boreczky. Flyspec: A multi-user video camera system with yspec: A multi-user video camera system with hybrid human and automatic control. International Multimedia Conference Proceedings of the tenth ACM international conference on Multimedia, 2002. 21
[15] F. Z. Qureshi and D. Terzopoulos. Surveillance camera scheduling: a virtual vision approach. ACM Multimedia Systems Journal, 2006. 20
[16] C. Stauffer and W. Grimson. Adaptive background mixture models for real-time tracking. CVPR, 1999. 43
[17] Wikipedia. Kalman lter. URL http://en.wikipedia.org/wiki/Kalman_filter. 10
[18] Z. Zhang. A exible new technique for camera calibration. PAMI, 2000. 7
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top