研究生: |
王文祺 Wang, Wen-chi |
---|---|
論文名稱: |
利用身高及人臉特徵實現遠距環境之數位家庭成員辨識系統 Far-Field People Identification for Family Members Using Height and Facial Features |
指導教授: |
王駿發
Wang, Jhing-fa |
學位類別: |
碩士 Master |
系所名稱: |
電機資訊學院 - 電機工程學系 Department of Electrical Engineering |
論文出版年: | 2009 |
畢業學年度: | 97 |
語文別: | 英文 |
論文頁數: | 53 |
中文關鍵詞: | 身高測量 、單視點量測 、人物識別 、遠距離 |
外文關鍵詞: | face recognition, far-field, height measurement, single-view metrology |
相關次數: | 點閱:71 下載:4 |
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 |
隨著生物特徵科技發展趨於成熟,其在許多「安全防護」上的應用日益增加,電影中常常利用虹膜、指紋、聲音或者是人臉來辨識某個人的身份,不過受限於這些生物特徵的特性與辨識率,市面上辨識系統產品發展方向常著重於近距離。我們不禁思考,是否有拓展其應用環境的可能性。
我們都知道隨著辨識距離越遠,其生物特徵也越來越模糊,這對於仰賴於特徵明顯程度的辨識率也隨之下降。因此是否可以結合遠距離的生物特徵(如:身高),提高當受限於應用環境時降低的辨識率。本論文實現了兩個生物特徵辨識系統,第一種是近距離特徵--人臉,利用FLD(Fisher Linear Discriminant)演算法,FLD是一種目前廣為人使用的方法,先利用PCA(Principle Component Analysis)降低一張影像的維度,藉此減少運算量,再利用LDA(Linear Discriminant Analysis)擷取特徵值,經由LDA運算,可以降低不同光線、表情等因素對於辨識率的影響。
第二種是遠距離特徵--身高,當考慮如何使用較低運算量與較低成本的情況下,我們選用「單視點量測(Single-View Metrology)」的方式。首先,校正因鏡頭失真的影像,然後給定一個在環境中我們已知的高度參考點,藉由此參考點我們可以計算出待測對象的身高。
本系統將會計算待測人物遠、近距離生物特徵,再將此兩特徵做交集運算,改善人臉特徵不明顯的缺陷,藉此提高人物辨識度,實驗結果顯示運用此系統能夠有效增加遠距離環境人物辨識率。
The development of biometric technology tends to be more and more maturative, therefore the ever-increasing applications for security have become more accessible to people. It is usually to identify a person with iris, fingerprints, voice and face in the movies. Now, most of the recognition systems are focused on short-distance applications by the characteristic restriction of biological features. Hence, it may be considered that if there is any possibility to extend the application field.
In far-field conditions, the recognition rate would be decreased because the biological features get more and more blurred. It is possible to enhance recognition rate by combining a far-field biological features such as height in a limited application environment. This thesis uses two biological features to develop identification algorithms. The first biological feature used in this thesis is face and it is done by FLD (Fisher Linear Discriminant) algorithm. The FLD is a famous algorithm for face recognition and is composed by two components. The first component of the FLD algorithm is using PCA (Principle Component Analysis) to reduce the dimension of an image and to reduce a computational complexity. LDA (Linear Discriminant Analysis) is the second part of FLD which is used to extract the features of image. After LDA, it can decrease the influence factor of recognition rate caused by different lightings and expressions.
The second biological feature is height. The height measurement algorithm is using the single-view metrology when considering a purpose of low cost and low computational complexity. The first step of the algorithm is to process the camera distortion. The second is to set the known reference point of height in the environment. The height of objects will be calculated by using this reference point.
This system will calculate both a long distance feature (height) and a short distance feature (face), and the interaction of the two features can improve the disadvantage caused by the blurred features. Hence the recognition rate will be increased. In this thesis, The experimental results show that the recognition rate in long distance environment can be improved by the proposed system.
[1] Lee, W.S.; Lee, H.J.; Chung, J.H., “Wavelet-based FLD for face recognition,” Proceedings of the 43rd IEEE Midwest Symposium on Circuits and Systems, 2000, , vol.2, pp.734-737.
[2] M. Turk and A. Pentland, “Eigenfaces for Recognition,” J. Cognitive Neuroscience, vol. 3, no. 1, 1991.
[3] M. Turk and A. Pentland, “Face Recognition Using Eigenfaces,” Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 1991, pp. 586-591.
[4] D.L. Swets and J.J. Weng, “Using Discriminant Eigenfeatures for Image Retrieval,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 18, no. 8, pp. 831-836, Aug. 1996.
[5] Aleix M. Martinez, and Avinash C. Kak, "PCA versus LDA", IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23,pp.228-233, February 2001.
[6] Belhumeur, P.N.; Hespanha, J.P.; Kriegman, D.J., "Eigenfaces vs. Fisherfaces: recognition using class specific linear projection," Pattern Analysis and Machine Intelligence, IEEE Transactions on , vol.19, no.7, pp.711-720, Jul 1997.
[7] David. G. Lowe. “Distinctive image features from scale-invariant keypoints,” International Journal of Computer Vision,2004, 60(2):91-110.
[8] Liu, C. and Wechsler, H.(2002): Gabor Feature Based Classification Using the Enhanced Fisher Linear Discriminant Model for Face Recognition, IEEE Transaction on Image Processing, Vol. 11(4), pp. 467-476.
[9] M. Grabner, H. Grabner, and H. Bischof. Fast approximated SIFT. In ACCV, 2006.
[10] Y. Ke and R. Sukthankar. PCA-SIFT: A more distinctive representation for local image descriptors. In Proceedings of the Conference on Computer Vision and Pattern Recognition, Washington, USA, pages 511. 517, 2004.
[11] K Fukunaga, “Introduction to Statistical Pattern Recognition”, Academic Press, second edition, 1991.
[12] Z.Q. Hong, and J.Y. Yang, “Optimal discriminant plane for a small number of samples and design method of classifier on the plane”, Pattern Recognition, Vol.24, no.4, pp.317-324, 1991.
[13] Chengjun Liu; Wechsler, H., "Enhanced Fisher linear discriminant models for face recognition," Pattern Recognition, 1998. Proceedings. Fourteenth International Conference on , vol.2, no., pp.1368-1372 vol.2, 16-20 Aug 1998
[14] Bulent Atalay, “Math and the Mona Lisa: The Art and Science of Leonardo da Vinci”, (2007).
[15] A. Criminisi, I. Reid and A. Zisserman: Single View Metrology. Proceedings of the 7th International Conference on Computer Vision (ICCV’99), September 1999, Kerkyra, Greece, pp. 434-441.
[16] A. Criminisi, Accurate Visual Metrology from Single and Multiply Images. Distingiushed Dissertation Series. Springer-Verlag London Ltd., Sept. 2001.
[17] A. Criminisi, “Single-view metrology: Algorithms and applications,” in DAGM-Symposium, Zurich, Sept. 2002, pp. 224–239.
[18] Leon Battista Alberti. De Pictura. 1435. Reproduced by Laterza (1980).
[19] Richard Hartley and Andrew Zisserman, “Multiple View Geometry in Computer Vision”, Cambridge University Press.
[20] C. E. Springer, ”Geometry and Analysis of Projective Spaces”, Freeman, 1964.
[21] F. Schaffalitzky and A. Zisserman. Planar grouping for automatic detection of vanishing lines and points. Image and Vision Computing, 18:647–658, 2000.
[22] Murphy-Chutorian, E., & Trivedi, M. M. (2008). Head pose estimation in computer vision: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence (in press).
[23] P. Viola, M. Jones. Rapid object detection using a boosted cascade of simple features. IEEE Conf on Computer Vision and Pattern Recognition, Kauai, Hawaii, USA: IEEE Computer Society, 2001.