進階搜尋


 
系統識別號 U0026-0812200914021173
論文名稱(中文) 使用多台相機的三維手指影像運動分析系統
論文名稱(英文) Three-Dimensional Finger Motion Analysis System Using Videos from Multiple Cameras
校院名稱 成功大學
系所名稱(中) 資訊工程學系碩博士班
系所名稱(英) Institute of Computer Science and Information Engineering
學年度 95
學期 2
出版年 96
研究生(中文) 何聖斌
研究生(英文) Sheng-Pin Ho
電子信箱 p7694403@mail.ncku.edu.tw
學號 p7694403
學位類別 碩士
語文別 中文
論文頁數 122頁
口試委員 指導教授-孫永年
口試委員-蔡清欉
口試委員-蘇芳慶
口試委員-陳澤生
口試委員-郭立杰
中文關鍵字 粒子濾波器  運動分析系統  手指運動學  多台相機  三維模型 
英文關鍵字 motion analysis system  particle filtering  kinematics  model  multiple cameras 
學科別分類
中文摘要 依據醫學上的研究,三維手部動作的分析與量測對手部疾病之診斷與治療能提供很大的幫助。在臨床上,醫師以動態螢光攝影X光影像診斷手部病變。但是動態X光攝影只提供二維的投影資訊,對於提供整體手傷運動之評估與診療仍有不足。有鑒於此,本系統藉由電腦視覺技術與三維虛擬手指模型,發展一套三維手指動作分析與量測系統。然而手部動作具有高自由度及限制性的特性,常常會因為姿態與動作的不同,而產生遮蔽現象。本論文使用多台相機以降低遮蔽效應,並藉以增加量測手指運動參數的精確性與實用性。
目前部分研究使用追蹤標記點的三維運動分析系統,然而標記點所提供的資訊有限,無法表現手指的結構與自由度。我們的系統整合標記點以預測模型參數,同時使用手部影像以調整改善模型參數,所以必須同時追蹤標記點和分析輸入手部影像。本論文主要分為兩部份:預測標記點三維資訊,以及預測模型參數。在預測標記點三維資訊部分,我們藉由電腦視覺之極點限制和空間幾何限制,解決了多台相機之複雜對應問題。同時也改善了傳統的Mean-Shift追蹤演算法以防止標記點在時間序列上產生過大的跳動現象。最後提出了一套自動選擇視野的重構方法,以排除不可靠的視野。經過實驗證實本系統之重構誤差均較使用兩台相機者為低。
在預測模型參數部份,我們改善傳統的粒子濾波器演算法,藉由動態整合由標記點定義的模型參數與前ㄧ時間點的最佳參數,使權重較大的粒子能夠有效的被取樣出來。為增加系統的可靠度,故藉由標記點資訊改善了粒子濾波器的Weighting Function,並加入了限制項來濾除不合理的參數。本論文最後會將MRI所得到的三維手骨模型放入所建構之虛擬指塊模型中,以顯示模擬手傷患者在動作時骨骼之間的運動狀況與旋轉角度等。
由實驗結果顯示,本論文在單指的食指彎曲動作,多指的對指動作與全指的球狀抓握動作等都有良好的追蹤結果。藉由本系統可以量測到不同手指的運動參數,例如骨節與骨節之間的旋轉角度、交錯位移,骨節的長度,骨節的座標軸,運動軌跡,角速度與角加速度等資訊,同時藉由多台相機系統也大大降低了遮蔽效應的現象。
本論文提供了一個動態分析手指運動的系統,利用此系統,醫生可以方便的記錄病患術前術後的動態資訊,進而建立完整的手部動態病例。
英文摘要 In hand disease researches, 3D motion analysis provides valuable information to doctors for the diagnosis and treatment of hand functional disorder. Recently, the real-time fluoroscopy analysis is used to diagnose the underlying condition of hand-injured patients, it provides dynamic 2D information which seems insufficient for the evaluation of 3D hand motion. Therefore, we utilize computer vision techniques to build the proposed 3D virtual hand model and to set up an integrated 3D hand motion analysis system. Since the finger motion is with high degrees of freedom and also highly constrained, the complex occlusion problem of finger motion may arise. In the proposed system, four cameras were used to solve the occlusion problem and to increase the system stability and the accuracy of finger motion parameters.
The conventional gait analysis is purely based on marker tracking to measure hand motion parameters. However, the information provided by the markers is limited and cannot provide the structural information of real hand, e.g. the transformation between bone segments. In the proposed model-based system, the 3D marker information is used to predict the model and the image information is used to refine the model parameters. Therefore, our approach consists of two important components: detecting the 3D information of markers and predicting the model parameters. For the former, we utilize the epipolar lines and the geometric constraint in computer vision to solve the complicated correspondence problem. We also modified the traditional mean-shift algorithm to prevent the severe marker jumps in time sequence. Finally, we proposed an automatic view selection method to eliminate outlier views. The experiments show the proposed method achieves less reconstruction errors than the ones obtained by the reconstruction from two cameras. For the latter, we improved the traditional particle filtering algorithm by adjusting the particle sampling process, which dynamically integrates the model parameters defined by markers and the optimal model parameters predicted in the previous frame. In order to increase the system stability, we improved the weighting function via marker information and added the constraint term to eliminate invalid parameters.
Finally, the proposed system also employs the bone mesh reconstructed from MRI and integrates the mesh into the 3D virtual hand model such that we can observe the movement and rotation angles between bone segments during the hand motion of patient.
The proposed system achieves good tracking results in the single finger motion (index F/E), multiple fingers motion (pinch), and all fingers motion (spherical grasp) experiments. It can measure the motion parameters for different fingers, including rotation angles and translation between bone segments, local axis of bone segments, motion trajectory, angle velocity, and angle acceleration. It also lessens the occlusion problem by using multiple cameras.
This thesis proposed a dynamic finger motion analysis system. The medical doctors can record the hand motion information for both pre- and post-operative evaluations of patient conveniently.
論文目次 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 6
第二章、理論背景 8
2-1.建立以手指運動學為基礎之參考手指模型 8
2-1-1.虛擬三維手指模型介紹 8
2-1-2.模型參數介紹 12
2-2.極點幾何理論 14
第三章、研究方法 19
3-1.影像特徵擷取 19
3-1-1.手指及標記點前景分割 19
3-1-2.手指輪廓萃取 23
3-2.模型參數初始化 24
3-3.預測三維標記點資訊 32
3-3-1.標記點偵測與對應 32
3-3-2.標記點追蹤 39
3-3-3.重構標記點三維資訊 45
3-4.預測三維模型參數 50
3-4-1.由標記點定義模型參數 51
3-4-2.整合標記點資訊的適應性粒子濾波器 54
第四章、實驗設計與結果討論 66
4-1.受測者的前置作業 66
4-2.受測者的手指動作 67
4-3.相機校正 68
4-4.效能實驗 71
4-4-1.效能實驗環境 71
4-4-2.效能實驗結果 73
4-4-3.Modified Mean-Shift和Classic Mean-Shift的比較 86
4-4-4.多台相機與兩台相機重構誤差比較 89
4-4-5.Adaptive Marker-Guided Particle Filtering 和Classic Particle Filtering的比較 92
4-5.驗證實驗 95
4-5-1.驗證實驗環境 95
4-5-2.驗證實驗設計 96
4-5-3.驗證實驗步驟 99
4-5-4.驗證實驗結果 102
4-6.運動參數分析 104
第五章、結論與未來展望 115
5-1. 結論 115
5-2. 未來展望 116
參考文獻 118
參考文獻 [1] 陳彥廷、孫永年,”以模型為基礎之電腦手指連續運動影像分析系統”,國立成功大學資訊工程研究所碩士論文,2006。
[2] X. Zhang, S.-W. Lee and P. Braido, “Determining finger segmental centers of rotation in flexion–extension based on surface markers measurement,” Journal of Biomechanics, vol. 36, iss. 8, pp. 1097-1102, August 2003.
[3] P. Cerveri, N. Lopomo, A. Pedotti, and G. Ferrigno, “Derivation of centers and axes of rotation for wrist and fingers in a hand kinematic model: methods and reliability results,” Annals of Biomedical Engineering, vol. 33, no. 3, pp. 402-412, March 2005.
[4] G.S. Rash, P.P. Belliappa, M.P. Wachowiak, N.N. Somia, and A. Gupta, “A demonstration of the validity of a 3-D video motion analysis method for measuring finger flexion and extension,” Journal of Biomechanics, vol. 32, iss. 12, pp. 1337-1341, December 1999.
[5] L.-C. Kuo, F.-C. Su, H.-Y. Chiu, and C.-Y. Yu, “Feasibility of using a video-based motion analysis system for measuring thumb kinematics,” Journal of Biomechanics, vol. 35, iss. 11, pp. 1499-1506, November 2002.
[6] F.-C. Su, L.-C. Kuo, H.-Y. Chiu, and H.-Y. Hsu, “The validity of using a video-based motion analysis system for measuring maximal area of fingertip motion and angular variation,” Proceedings of the Institution of Mechanical Engineers. Part H, Journal of Engineering in Medicine, vol. 216, no. 4, pp. 257-263, July 2002.
[7] F.-C. Su, L.-C. Kuo, H.-Y. Chiu, and M.-J. Chen-Sea, “Video-computer quantitative evaluation of thumb function using workspace of the thumb,” Journal of Biomechanics, vol. 36, iss. 7, pp. 937-942, July 2003.
[8] Shan Lu, D. Metaxas, D. Samaras and J. Oliensis, “Using multiple cues for hand tracking and model refinement”, IEEE Conf. on Computer Vision and Pattern Recognition 2003, 18-20 Jun. 2003, vol.2, pp. 443-450.
[9] R. Rosales, V. Athitsos, L. Sigal, and S. Sclaroff, "3D Hand Pose Reconstruction Using Specialized Mappings" IEEE ICCV. Canada. Jul. 2001.
[10] Arulampalam, M.S. Maskell, S. Gordon, N. Clapp, T. ,“A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking”, Signal Processing, IEEE Transactions , vol. 50, iss.2, pp. 174-188, Feb 2002.
[11] Y. Wu, J. Lin and T. Huang, “Analyzing and capturing articulated hand motion in image sequences”, IEEE Trans. PAMI 27(12) (2005) 1910–1922.
[12] Bray, M., Koller-Meier, E., and Van Gool, L., “Smart particle filtering for 3D hand tracking”, . Int. Conf. on Automatic Face and Gesture Recognition, 2004, pp. 675–680.
[13] J. Lin, Y. Wu, and T.S. Huang, “Modeling the constraints of human hand motion,” in Proc. of IEEE Human Motion Workshop, 2000, pp. 121-126.
[14] J. Lin, Y. Wu, and T.S. Huang, “Modeling the natural hand motion constraints,” in Proc. of ARL Federated Laboratory 5th Annual Symposium, 2001. pp. 105-110.
[15] CY Chien, CM Fu, CY Weng, CL Huang, “Vision-Based Real-Time 3-D Arm Pointing System Using Multiple Cameras”, Proc. Conf. Computer Vision, Graphics and Image Processing, 2006.
[16] B Liu, R Manner, “An Efficient and Accurate Method for 3D-Point Reconstruction from Multiple Views“, International Journal of Computer Vision 65(3), 175–188, 2005.
[17] C.-S. Chua, H. Guan, and Y.-K. Ho, “Model-based 3D hand posture estimation from a single 2D image,” Image and Vision Computing, vol. 20, iss. 3, pp. 191-202, March 2002.
[18] J. Lin, Y. Wu, and T.S. Huang, “Capturing human hand motion in image sequences,” in Proc. of IEEE Workshop on Motion and Video Computing, 2002, pp. 99-104.
[19] P. W. Brand and A. Hollister, Clinical Mechanics of the Hand, 2nd ed.,St. Louis, MO: Mosby Year Book, 1993.
[20] E Trucco, A Verri. Introductory Techniques for 3-D Computer Vision. Upper Saddle River, New Jersey: Prentice Hall, 1998
[21] Kanade T, Okutomi M. A stereo matching algorithm with an adaptive window: Theory and Experiments. IEEE Trans, on PAMI. 1994,16(9) :920~932.
[22] Mori K., Kidode M. and Asada H., “An iterative prediction and correction method for automatic stereo comparison,” Computer Graphics and Image Processing, 2:393-401, 1973.
[23] N. Ayache and B. Faverjon, "Efficient registration of stereo images by matching graph descriptions of edge segments," International Journal of Computer Vision, pp. 107-131, 1987.
[24] W.E.L. Grimson. "Computational experiments with a feature based stereo algorithm," M.I.T., Cambridge,MA, A.I. Memo. 762, 1984.
[25] ZY.Zhang, “Determining the epipolar geometry and its uncertainty: a review,” The International Journal of Computer Vision, 1998, 27
[26] D. Comaniciu, V. Ramesh, and P. Meer, “Kernel-Based Object Tracking,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 25, no. 5, pp. 564-575, May 2003.
[27] Isard,M.,Blake,A.,”Condensation - conditional density propagation for visual tracking,”International Journal on Computer Vision,1998,29(1):5-28.
[28] A. Doucet, S. Godsill, and C. Andrieu, “On sequential Monte Carlo sampling methods for Bayesian filtering,” Statist. Comput., Vol. 10, No. 3, pp. 197-208, 2000.
[29] V.I. Pavlovic, R. Sharma, and T.S. Huang. “Visual interpretation of hand gestures for human-computer interaction: a review,” IEEE Transaction on Pattern Analysis and Machine Intelligence, Vol.19, No.7, July 1997.
[30] G. Kitagawa, “Monte Carlo filter and smoother for non-Gaussian nonlinear state space models,” J. Comput. Graph. Statist., vol. 5, no. 1, pp. 1–25, 1996.
[31] Z. Zhang, “A flexible new technique for camera calibration,” IEEE Trans. Pattern Anal. Machine Intell., vol. 22, no. 11, pp.1330–1334, Nov. 2000.
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2009-09-10起公開。
  • 同意授權校外瀏覽/列印電子全文服務,於2009-09-10起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw