簡易檢索 / 詳目顯示

研究生: 楊上毅
Yang, Shang-Yi
論文名稱: 人臉追蹤與辨識之自動化視訊監控系統
Automated surveillance system for face tracking and recognition
指導教授: 孫永年
Sun, Yung-Nien
學位類別: 碩士
Master
系所名稱: 電機資訊學院 - 資訊工程學系
Department of Computer Science and Information Engineering
論文出版年: 2008
畢業學年度: 96
語文別: 中文
論文頁數: 90
中文關鍵詞: 人臉追蹤人臉辨識
外文關鍵詞: people recognition, people tracking
相關次數: 點閱:56下載:3
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 現今視訊監控系統已經廣泛地應用在門禁管制與交通監測等方面。但多數系統僅具備簡易的資料記錄與分析功能,常需專人介入主控監視。對於公務或軍事等重要機構或危險人物的管制,顯然已經無法滿足現在的需求了。有鑑於此,本論文提出一套以靜態與動態攝影機為基礎之雙視訊監控系統,可用於自動臉部偵測,追蹤與辨識。此技術利用光學攝影機在固定監視範圍內,進行移動物體之偵測,接著PTZ 攝影機進行人物之拍攝與追蹤,最後辨識人物的身分。
    本論文主要有三個重點: 人臉偵測、人臉追蹤和人臉辨識。在人臉偵測方面,靜態攝影機會在監視範圍內進行監控,利用背景減除的方式得知有無人物進入監視範圍。一旦偵測出有人物出現,我們利用影像前處理將人物影像轉換成FPI,依據FPI 的資訊找出屬於人臉的部分。接下來透過極線的方式將人臉座標參數傳遞給動態攝影機。
    在人臉追蹤方面,利用FPI 的資訊,使用Camshift 進行人臉追蹤。因為人物會移動,所以動態攝影機會依據人物的位置來轉動鏡頭,保持人物能夠持續出現在畫面中。追蹤過程中若是喪失膚色資訊,將會造成追蹤失敗。此情況一旦發生,我們會啟動補償機制來重新追蹤目標物。
    在人臉辨識方面,我們會先建立視覺系統實驗室的人臉資料庫。採用2DPCA,可以取出這些人臉資料的主成份。用這些主成分來建立多棵kd-tree,這將有助於之後的人臉辨識。採用的系統具即時性,因此在追蹤人臉的同時會擷取人臉影像來辨識其身分。辨識的方法是將一張需要辨識的人臉影像放進kd-tree演算法,結合多棵kd-tree 所得到的結果,來判斷此人物的身份。此外,我們輸入多張人臉影像到辨識系統,以提高系統辨識的穩定性。經由實驗結果顯示出,本論文所提出的方法能夠正確的偵測、追蹤和辨識目標物。

    Surveillance systems have been popularly applied to visitor control, traffic monitor and many other fields nowadays. Most of them only have simple functions such as data records together with simple information analyses, and user interventions are usually inevitable. Thus, traditional systems depending on manual management can not satisfy the demands in many applications. To solve these problems, we propose a double-camera (a static and a dynamic) surveillance system that can track human and recognize face automatically. The proposed method utilizes the static camera (CCD) to monitor a fixed field of view, and detect the moving object in that area. After detecting a moving subject, static camera can send parameters of human position to dynamic camera (a PTZ). It enables the PTZ to track people and then extract the zoom-in face images. Finally, the proposed system can recognize the
    identity of the tracked people.
    There are three topics in the thesis. They are face detection, face tracking, and the people recognition. In face detection, static camera monitors the desired field of
    view, and we then use background subtraction method to detect whether a human enters. Once a human shows, we transform the detected human image into flesh probability image (FPI) and then find his/her face based on FPI information. Subsequently, the system sends parameters of face coordinates to the dynamic camera based on epipolar line geometry.
    In face tracking, we use Camshift algorithm relied on the FPI information to track people. Because human walks everywhere, the dynamic camera pans to the human direction according to his/her position to keep him/her in the middle of screen. In tracking processing, if the system losses skin color information, the tracking task may fail. Once the situation occurred, the system will start a compensation strategy to track people again.
    In people recognition, a face database is constructed from Visual System Laboratory. By using 2DPCA, it can extract principal components of these face images. Based on these principal components, we then construct kd-trees which are beneficial to the subsequent human recognition. The proposed system handles a real time procedure, thus when it can track a subject and obtain its face for recognition. The method of recognition is to input a new face image into the kd-tree algorithm, it is recognized by combining the search results from multiple kd-trees. Moreover, to enhance the robustness of the system, multiple face images are used in the recognition process. In the experimental study, the proposed method can detect, track, and recognize people successfully.

    摘要 I Abstract III 目錄 i 表目錄 iii 圖目錄 iv 第一章、序論 1 1-1 研究動機 1 1-2 相關研究 2 1-3 論文架構 4 第二章、動、靜態攝影機參數校正與建立對應性 7 2-1 相機參數校正 7 2-2 建立對應性 8 2-2-1 計算極線方程式 9 2-2-2 建立不同場景的對應性 11 第三章、人臉偵測與追蹤 22 3-1 影像前處理 22 3-2 人臉偵測 24 3-2-1 影像邊緣偵測 24 3-2-2 影像前景的求取 26 3-2-3 計算前景目標物臉部的質心 28 3-3 質心對應到動態攝影機 31 3-3-1 動態攝影機臉部的質心位置 32 3-4 人臉追蹤 34 3-4-1 使用Camshift的人臉追蹤 34 3-4-2 在推測的搜尋範圍下進行追蹤 35 3-4-3 動態攝影機的控制與質心座標的更新 37 3-5 追蹤失敗的補償機制 37 第四章、臉部辨識 40 4-1 取得臉部影像 40 4-2 2DPCA 42 4-2-1 基本概念與演算法 43 4-2-2 特徵提取 (Feature Extraction) 45 4-2-3 以2DPCA為基礎的影像重構 46 4-2-4 辨識方法 48 4-3 多棵kd-tree進行辨識 50 4-3-1 Kd-tree概念 50 4-3-2 建立多棵kd-tree 53 4-3-3 結合多棵kd-tree的結果 55 第五章、實驗結果與討論 60 5-1 實驗環境 60 5-2 實驗結果 62 5-2-1 FPI的製作 62 5-2-2 使用極線傳遞目標物臉部座標 63 5-2-3 人臉追蹤 65 5-2-4 Vision_Lab database的建立 67 5-2-5 計算不同kd-tree所需的鄰居數 68 5-2-6 比較使用kd-tree和使用最近鄰居分類器的運算時間 69 5-2-7 偵測、追蹤和辨識的花費時間 70 5-2-8 人臉辨識 72 5-3 討論 81 第六章、結論與未來展望 84 6-1 結論 84 6-2 未來展望 86 參考文獻 87

    [1] C. Garcia and G. Tziritas, “Face detection using quantized skin color regions merging and wavelet packet analysis”, IEEE Trans. On Multimedia, Vol.1 No.3, pp. 264-277, 1999.
    [2] P. Hall, D. Marshall, and R. Martin, “Incremental eigenanalysis for classification,” In British Machine Vision Conference , Vol. 1, pp. 286-295, 1998.
    [3] M. Turk and A. Penland, “Eigenfaces for recognition,” Journal of Cognitive Neuroscience, Vol.3, pp.71-86, March 1991.
    [4] S.J. Leon, Linear Algebra with Application, Prentice Hall, 2002.
    [5] J.O. Rawlings, Applied Regression Analysis, Wadsworth & Brooks/Cole,Pacific Grove, CA, 1988.
    [6] N. Roy, G. Baltus, D. Fox, F. Grmperle, J. Gortz, and T. Hirsch, “Towards Personal Service Robots for the Elderly,” Workshop on Interactive Robots and Entertainment (WIRE), 2000.
    [7] R.L. Hsu, M. Abdel-Mottaleb, A.K. Jain, “Face detection in color images,” IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 24, pp. 696-706, 2002.
    [8] C. Garcia and G. Tziritas, “Face detection using quantized skin color regions merging and wavelet packet analysis,” IEEE Trans. Multimedia, Vol. 1 No. 3, pp. 264-277, 1999.
    [9] C.C. Han, H.Y. Liao, G.J. Yu, and L.H. Chen, “Fast face detection via morphology-based pre-processing,” Pattern Recognition, Vol. 33, pp. 1707-1712, 2000.
    [10] C.H. Lin and K.C. Fan, “Triangle-based approach to the detection of human face,” Pattern Recognition, Vol. 34, pp. 1271-1284, 2001.
    [11] S. Birchfield, “Elliptical head tracking using intensity gradients and color histograms,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.232-237, 1998.
    [12] L. Jordao, M. Perrone, J.P. Costeira, and J. Santos-Victor, “Active face and feature tracking,” Proceedings of International Conference on Image Analysis and Processing, pp. 572-576, 1999.
    [13] C. Wengert, T. Fong, S. Grange, and C. Baur, “Human-oriented tracking for human-robot interaction,” Int. Conf. on Multimodal Interfaces, 2002.
    [14] X. Xie, R. Sudhakar, and H. Zhuang, “A cascaded scheme for tracking and head movement compensation,” IEEE Trans. Systems, Man, and Cybernetics - Part A: Systems and Humans, Vol. 28, no. 4, pp. 487-490, July 1998.
    [15] A. Azarbayejani and A. Pentland, “Recursive estimation of motion, structure, and focal length,” IEEE Trans. Pattern Analysis and Machine Interlligence, Vol. 17, pp. 562-575, 1995.
    [16] Y. Cheng, “Mean shift, mode seeking, and clustering,” IEEE Trans. Pattern Analysis and Matching Intelligence, Vol. 17, no. 8, 1995.
    [17] R. Cendrillon, Real-Time Face Recognition using Eigenfaces, Master’s thesis, Department of Computer Science and Electrical Engineering Queensland University, 1999.
    [18] M.J. Er, S. Wu, j. Lu, and H. Lye, “Face Recognition with Radial Basis Function (RBF) Neural Networks,” IEEE Transactions on Neural Networks, Vol. 13, No. 3, pp.697-710, 2002.
    [19] P.N. Belhumeur, J.P. Hespanha, and D.J. Kriegman, “Eigenfaces vs. Fisherfaces : recognition using class specific linear projection,” IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19, No. 7, pp.711-720, 1997
    [20] K.M. Lam and H. Yan, “An analytic- to- holistic approach for face recognition based on a single frontal view,” IEEE Trans. On Pattern Analysis and Machine Intelligence, Vol. 20, No. 7, pp.673-686, 1998.
    [21] H. Stern and B. Efros, “Adaptive color space switching for tracking under varying illumination,” Image and Vision Computing, Vol. 23, pp. 353-364, 2005.
    [22] Z. Zhang, “A flexible technique for camera calibration,” IEEE Trans. Pattern Analysis and. Machine Intelligence, Vol. 22, No. 11, pp. 1330-1334, 2000.
    [23] G.R. Bradski and S. Clara, “Computer vision face tracking for use in perceptual use interface,” Intel Technology Journal, 1998.
    [24] Jian Yang, David Zhang, Alejandro F. Frangi, and Jing-yu Yang, “Two -Dimensional PCA: A New Approach to Appearance-Based Face Representation and Recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.26, No.1, pp.131-137, 2004.
    [25] Daoqiang Zhang, Zhi-Hua Zhou, “(2D)2PCA: 2-directional 2-dimensional PCA for efficient face representation and recognition,” Neurocomputing, Vol.69 , pp.224-231, 2005.
    [26] Jian Yang, Jing-yu Yang, “From image vector to matrix: a straightforward image projection technique—IMPCA vs. PCA,” Pattern Recognition, Vol.35, No.9, pp.1997-1999, 2002.
    [27] Xiao-Yuan Jing, Hau-SanWong, David Zhang, “Face recognition based on 2D Fisherface approach,” Pattern Recognition, Vol.39, pp.707-710, 2006.
    [28] D.M. Mount and S. Arya, “ANN: A Library for Approximate Nearest Neighbor Searching,” Proc. Center for Geometric Computing Second Ann. Fall Workshop Computational Geometry, 1997. (http://www.cs.umd.edu/~mountANN.)

    下載圖示 校內:2011-08-26公開
    校外:2013-08-26公開
    QR CODE