臺灣博碩士論文加值系統

自動化的環境監控系統，在近年來是一個相當熱門的主題，尤其是在電腦視覺領域上，以影像辨識為基礎用在自動環境監控系統上更是活躍。一般監控系統大抵只為了達到目標物的偵測與追蹤兩個目的，我們希望能結合全景虛擬實境的概念，將一個全景影像與可轉動的攝影機結合，來建構一個自動化監控系統，以改善一般監控系統不易操作的缺點。
首先建立全景影像方面，我們利用比對影像中相對特徵點的方式來評估相機焦距長度，接著依影像的相對旋轉角度，將每張影像投影到一個半球面上，最後使用繪製地球地圖的方法將全景影像顯示在平面上，完成全景影像的建構。
就偵測追蹤前景物上，採用連續影像差異來偵測移動物體，並以臨近編碼統計來去除雜訊干擾，再利用分水嶺分割法，對目標物存在的部分影像區域進行影像分割，以取得目標物的影像資料，將移動物的質心利用卡曼濾波器，預測下一個時間點的目標物大略位置座標，最後透過預測的座標控制旋轉平台，讓移動物不會離開攝影機視線範圍，來完成追蹤移動物的目的。
我們實際實驗了戶外與室內的兩組狀況，實驗的結果顯示出系統的可行性。

Recently, video surveillance and monitoring (VSAM) has gradually become a popular research topic due to its importance. In such systems, the main tasks to be performed can be summarized as follows; image sequences are processed, moving targets are detected and tracked, video data are stored and queried, and alarms are made when illegal events occur. Many topics focus on the study of image processing, pattern recognition, and artificial intelligence.
In this thesis, an immersed environment is developed by combining virtual reality and video monitoring methodologies. The image sequences are grabbed to construct the panoramic images from a programmable controlled PTZ camera. These images are stitched by using the moisack techniques. In addition, moving targets are detected, tracked and segmented from the video sequences. Moreover, the PTZ camera is triggered by the predicted data in order to continuously capture the images with moving targets when objects are out of the field of view (FOV). Two experimental environments, an indoor scene and an outdoor scene, are constructed to demonstrate the validity and effectiveness of our proposed approach.

Abstract i
摘要 ii
目錄 iii
附圖目錄 vi
第一章 緒論 1
1.1 研究動機 1
1.2 相關研究 2
1.3 系統流程 6
1.4 論文架構 8
第二章 建立全景影像 9
2.1 系統硬體模型 9
2.2評估焦距長度 10
2.2.1：透視投影轉換 11
2.2.2：二維投影轉換 12
2.2.3：特徵點選取 16
2.2.4：計算轉換矩陣 17
2.3影像接合 19
2.3.1：俯仰角計算 20
2.3.2：水平角計算 22
2.4麥卡托投影(Mercator Projection) 24
第三章 偵測與追蹤 25
3.1 移動物偵測 25
3.1.1：鄰近點編碼去雜訊 28
3.1.2：決定前景範圍 29
3.2 移動物追蹤 30
3.2.1：分水嶺分割法 31
3.2.2：卡曼濾波器 34
3.3全景影像顯示移動物 40
第四章 實驗結果 41
4.1 室外實驗結果 41
4.2 室內實驗結果 47
第五章 結論與未來研究方向 52
5.1 結論 52
5.2未來研究方向 53
參考文獻 54

[1] B. Hu, C. Brown, and A. Choi, “Acquiring an environment map through image mosaicking,”tech. rep., Computer Science Department, University of Rochester, 2001.
[2] S. Coorg and S. Teller, “Spherical mosaic with quaternions and dense correlation,” International Journal of Computer Vision, vol. 37,pp. 259–273, 2000.
[3] E. Noirfalise, J. Lapreste, F. Jurie, and M. Dhome, “Real-time registration for image mosaicing,” in Proc. of British Machine Vision Conference, 2002.
[4] F. Jurie and M. Dhome, “Real time template matching,” in Proc. IEEE International Conference on Computer Vision, 2001.
[5] G. L. Foresti and C. Micheloni, “Real-time video-surveillance by an active camera,” in Proc. of Workshop sulla Percezione e Visione nelle Macchine, Universita di Siena, 2002.
[6] R. Gupta, M. D.Theys, , and H. J. Siegel, “Background compensation and active-camera motion tracking algorithm,” in Proc. of International Conference on Parallel Processing, pp. 11–15,1997.
[7] N. Paragios and R. Deriche, “Geodesic active contours and level sets for the detection and tracking of moving objects,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 22, pp. 266–280, 2000.
[8] L. Vincent and P. Soille, “Watersheds in digital spaces: An efficient algorithm based on immersion simulations,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 13, pp. 583–598, 1991.
[9] S. Beucher, “The watershed transformation applied to image segmentation,”in Proc. Of Pfefferkorn Conference on Signal and Image Processing in Microscopy and Microanalysis,pp. 299–314, 1991.
[10] P. D. Smet, R. Luis, and V. P. M. Pires,“Implementation and analysis of an optimized rainfalling watershed algorithm,” in Proc. of Science and Technology Conference: Image and Video Communications and Processing, 2000.
[11] S. E. Hernandez and K. E. Barner, “Joint region merging criteria for watershed-based image segmentation,”in IEEE International Conference on Image Processing, pp. 108–111, 2000.
[12] T. Geraud, P. Y. Strub, and J. Darbon, “Color image segmentation based on automatic morphological clustering,” in Proc. of IEEE International Conference on Image Processing, pp. 70–73, 2001.
[13] R. Hartley, “In defense of the eight-point algorithm,”IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 19, pp. 580–593, 1997.
[14] E. W. Weisstein, “Mercator projection.”From MathWorld–A Wolfram Web Resource, http://mathworld.wolfram.com/Mercator Projection.html, 2004.
[15] N. S.Wu, “High altitude flying objects detection and tracking,” Master’s thesis, National Central University, 2003. (in Chinese).