臺灣博碩士論文加值系統

我們將會討論採取如何策略來估算空間的資訊，為了達到即時應用的最終目地我們將利用快速有效率的SUSAN找出空間中重要特徵點，並先建立簡單的對應關係，因為對應關係的求取在電腦視覺中是一個十分困難的問題，所以我們求取估來的對應關係相當的不穩定，所以我們將採用Zhang所提出的方法： 利用強健的方法篩除不正確的對應關係，並建立兩攝影機擷取的左右兩影像平面的極線幾何關係。接下來的問題便是如何從影像平面的極線幾何關係中求取空間中兩攝影機的相對應關係：旋轉及位移，要解決這個問題之前必須先對兩攝影機分別作自我校正的動作，因為兩部攝影機都具有可自我旋轉轉的功能，所以我們可以透過攝影機的自我旋轉來達到校正的目的，我們將會考慮攝影機內部參數固定與不固定的情況，最後我們可以透過攝影機的內部參數及外部參數，求出兩攝影的的相對關係，並重建兩相機的投影矩陣，藉此重構出整個三維空間場景。

We will discuss how to estimate the Euclidean space information. In order to achieve the ultimate goal of real-time applications, we will use a fast and efficient algorithm SUSAN to extract the important features of the targets. The matching problem is important and difficult in computer vision science. We will use a robust technique to remove the outliers and recover the epipolar geometry from two uncalibrated cameras. Furthermore, we will discuss how to estimate the relationship of two cameras (rotation and translation). To solve this problem, the cameras must perform self calibration first. So we consider two cameras which can be used thru self-rotation to estimate internal parameters. We will discus two cases: stationary internal parameters and varying internal parameters. Finally, we have internal parameters and external parameters. And we can estimate the Euclidean space information by reconstructing Euclidean space project matrix.

中文摘要………………………………………………………………………….……….I
英文摘要………………………………………………………………………………….II
目錄………………………………………………………………………………………III
圖目錄……………………………………………………..……………………………...V
第一章. 序論………………..…………………………..…………………………..…….1
1.1 簡介………………..……………………..…………………………..…….1
1.2 研究方法………………..…………………………..…………………..….2
1.3 章節概要………………..…………………………..…………………..….3
第二章. 成像幾何學及極線幾何………………..……………………………………….5
2.1 成像幾何學………………..……………………………………………….5
2.2 極線幾何限制………………..…………………………………………….7
2.2.1 必要矩陣………………..…………………………………………..9
2.2.2 基礎矩陣………………..…………………………………………..9
2.5 特徵點抽取………………..……………………………………………...10
2.5.1 SUSAN特徵抽取的主要原理………………………………..........11
2.5.2 SUSAN角點特徵抽取………………….………………………….12
2.6 特徵點對應……………………………………………………………….16
2.7 估算極線幾何…………………………………………………………….17
2.8 決定一個誤差函數……………………………………………………….18
2.9 強健的估算基礎矩陣…………………………………………………….19
第三章. 三維空間估測及攝影機自我校正…………………………………………….25
3.1 景深估算………………………………………………………………….25
3.2 攝影機自我校正………………………………………………………….29
3.2.1 實作攝影機自我校正……………………………………………..32
3.3 估計三維空間………………………………………………...…………..36
第五章. 實驗結果與討論………………………………………………...……………..39
4.1 系統環境………………………………………………...………………..39
4.2 特徵抽取穩定性分析………………………………………………...…..40
4.3 特徵點對應關係………………………………………………...………..42
4.4 估測極線幾何………………………………………………...…………..45
4.5 相機自我校正………………………………………………...…………..46
4.6 估測三維空間………………………………………………...…………..50
第六章. 結論與未來展望
5.1 結論………………………………………………...……………………..55
5.2 未來展望………………………………………………...………….…….55
參考文獻

[1] Tsai, R. and R. Lenz: 1989, 'A new technique for fully autonomous and efficient 3D robotics ''. hand/eye calibration''. IEEE Trans. Robotics and Automation 5(3), 345--358.
[2] Jianbo Shi and Carlo Tomasi., “Good Features to Track.”, IEEE Conference on Computer Vision and Pattern Recognition, pages 593-600, 1994.
[3] Stephen M. Smith and J.Michael Brady, “ SUSAN- A new approach to low level image processing”, International Journal of Computer Vision23(1) , 1997
[4] M. A. Fischler and R. C. Bolles. “Random sample consensus: A paradigm for model fitting with application to image anaysis and automated cartography.”, Comm. Assoc. Comp. Mach., 24(6):381-395, 1981
[5] Z. Zhang, R. Deriche, O. Faugeras, and Q. Luong., “A Robust Technique for Matching two Uncalibrated Images through the Recovery of the Unknown Epipolar Geometry”, Artificial Intelligence, 78:87-- 119, 1995.
[6] Z. Zhang, “Determining the Epipolar Geometry and its Uncertainty: A Review”, Technical Report RR-2927, INRIA, 1996
[7] Pollefeys, M.; Koch, R.; Van Gool, L., “Self-calibration and metric reconstruction in spite of varying and unknown internal camera parameters”, Computer Vision, 1998. Sixth International Conference on , 1998 , Page(s): 90 —95
[8] R. I.Harley.”self-calibration from multiple views with rotating camera”. In Proc. 3rd European Conf. on the Computer Vision, Stockholm, volume 1.pages 471-478,1994
[9] Hayman E., Agapito L., Reid I.D. and Murray D.W. (2000), “The role of self-calibration in Euclidean reconstruction from two rotating and zooming cameras”, in Proc 6th European Conf on Computer Vision, Vol II, Dublin, June/July 2000, pp 477--492
[10] Fadi Dornaika: “Self-Calibration of a Stereo Rig Using Monocular Epipolar Geometry”. ICCV 2001: 467-472
[11] W.Tiggs. Auto calibration and absolute quadric. In Proc. CVPR ,page 609-614,1997
[12] S.J. Maybamk and O.Faugeras. A theory of self-calibration of a moving camera”. International Journal of computer Vision ,8(2)123-151,1992
[13] R. I.Harley.”self-calibration of stationary cameras”. International Journal of computer Vision 22(1):5-23,Februaty 1997
[14] Z.Zhang, "A Flexible New Technique for Camera Calibration," Technical Report MSR-TR-98-71, Microsoft Research, Evil Empire, Redmond, WA, 98052, December 2, 1998 (see also http://research.microsoft.com/~zhang).
[15] Dornaika, F., Chung, R., “Cooperative Stereo-Motion: Matching and Reconstruction, CVIU(79)”, No. 3, September 2000, pp. 408-427
[16] Z. Zhang. “A new multistage approach to motion and structure estimation: From essential parameters to Euclidean motion via fundamental matrix”. Technical Report 2910, INRIA, 1996. 2
[17] X. Armangue, J. Pages, J. Salvi, J. Batlle. “Comparative Survey on Fundamental Matrix Estimation”. Simposium Nacional de Reconocimiento de Formas y Analisis de Imagenes, Castellon (Spain) May 2001
[18] P H S Torr and D W Murray, “The Development and Comparison of Robust Methods for Estimating the Fundamental Matrix” ,International Journal of Computer Vision Vol 24, No 3 (1997) pp271─300
[19] P. Beardsley, P. Torr and A. Zisserman, '3D Model Acquisition from Extended Image Sequences'' ECCV 1996.
[20] Richard Hartley, Andrew Zisserman, “Multiple View Geometry in Computer Vision”, CAMBRIDGE UNIVERSITY PRESS Publishing
[21] R.I. Hartley, “In Defense of the Eight-Point Algorithm”, IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19, No. 6, June 1997.
[22] Chien Ta-Yuan “Auto-Calibration, Reconstruction and Assessment of Clinical Lesions from Endoscopic Image Sequence”, Computer Science in information Engineering National Cheng-Kung University Taina, Taiwan, ROC