近年來，特效電影為電影工業的一大主流，這些特效場景必須利用影像合成的技術將演員合成至背景場面之中。為了合併出完美的合成影像，除了物件的亮度及顏色的強度需要經過校正外，另一個重點則是物件大小，根據置入的位置不同，物件的大小也要有所調整，因此需要利用圖片的深度來做大小的計算。因此本論文首先利用雙目立體視覺(Stereo Vision)和特徵點去計算出背景圖片的深度，接者針對物件的亮度及色彩強度做調整，並利用內插法配合深度資訊去做大小的調整，最後採用Possion演算法將校正過的物件融入背景之中。在本研究中，能夠有效利用深度資訊和背景資訊來對物件做校正，並且使用Possion editor來建構一套完整的影像編輯系統。

Over the last decade, the special effects movie is the most popular in the film industry. These special effects need to use the image combination to combine the avatars and the background. In order to get a perfect combination image, it has two key point: the illumination and the pixel intensity of the object and the object’s size. According to the different location, the object size also need to adjust. So that, it need the depth of the picture to measure the size. In our research, first, we use the stereo vision and the feature points to calculate the depth of the background. Second, to adjust the illumination and the pixel intensity of the object. Third, use the bilinear interpolation and the depth information to adjust the size. Finally, using the Possion algorithm to make the object melt in background naturally. In our research, we can use the depth and background information to adjust the object, and use the Possion editor to construct a complete video editing.

目錄
第一章	緒論	1
1.1.	研究背景	1
1.2.	研究動機與目的	2
1.3.	開發工具	4
1.4.	論文組織架構	7
第二章	相關研究工作	9
2.1.	立體視覺匹配(Stereo Matching)	9
2.2.	單目視覺(Monocular Cue)	14
2.3.	度量衡學(Metrology)	15
2.4.	全景式場景(Panorama)	17
第三章	研究方法與系統架構分析	19
3.1.	開發概念	19
3.2.	影像校正(Image Rectification)	20
3.2.1.	相機校正(Camera Calibration)	21
3.2.2.	同軸幾何(Epipolar Geometry)	26
3.2.3.	基礎矩陣(Fundamental Matrix)	28
3.2.4.	圖片校正	30
3.3.	特徵點比對	32
3.3.1.	影像積分法(Integral Images)	33
3.3.2.	Fast-Hessian Detector	35
3.3.3.	關注點描述器	38
3.4.	視差圖輸出	43
3.5.	全景化製作	45
3.6.	演員(Avatars)嵌入	46
3.6.1.	物件大小的計算	46
3.6.2.	根據背景資訊來對演員做調整	48
第四章	系統實作	54
4.1.	系統介面	54
4.2.	實驗結果	61
第五章	結論與未來展望	65
5.1.	結論	65
5.2.	未來展望	66
參考文獻	67
附錄 英文論文	72
 
圖目錄
圖 1、系統架構圖	19
圖 2、三大座標系統關係圖	22
圖 3、世界座標與相機座標關係圖	24
圖 4、校正用之棋盤圖片	25
圖 5、校正示意圖	26
圖 6、影像校正示意圖	26
圖 7、Epipolar Geometry	27
圖 8、本質矩陣的幾何關係	29
圖 9、影像積分法	34
圖 10、矩形面積總合(Summed Area Table, SAT)	35
圖 11、非最大值刪除法	38
圖 12、哈爾小波	39
圖 13、方位描述視窗	40
圖 14、描述器示意圖	41
圖 15、圖形特徵	41
圖 16、模糊特徵描述	42
圖 17、Sum of Absolute Difference	43
圖 18、SURF+SAD搜尋結果	43
圖 19、計算真實距離與視差值的樣本圖	44
圖 20、視差圖輸出	45
圖 21、演員樣本(一)	47
圖 22、演員樣本(二)	47
圖 23、沒有經過校正的範例	49
圖 24、BCB 6.0開發介面	54
圖 25、影像編輯器介面	55
圖 26、SURF選項	55
圖 27、特徵點搜尋結果	56
圖 28、視差圖輸出	57
圖 29、讀取背景資訊	57
圖 30、編輯介面	58
圖 31、設定速率及行走方向	58
圖 32、物件置入(一)	59
圖 33、物件置入(二)	59
圖 34、演員移動軌跡	60
圖 35、移動軌跡資訊	60
圖 36、置入結果預覽	61
圖 37、實驗結果(一)	62
圖 38、實驗結果(二)	62
圖 39、實驗結果(三)	63
圖 40、實驗結果(四)	63
圖 41、實驗結果(五)不同距離的大小差異	64

參考文獻
[1]	T. Kanade and M. Okutomi, “A Stereo Matching Algorithm with an Adaptive Window: Theory and Experiment,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 16, Issue. 9, pp. 920-932, 1994.
[2]	Y. Boykov, O. Veksler and R. Zabih, “Fast Approximate Energy Minimization via Graph Cuts,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, Issue. 11, pp. 1222-1239, 2001.
[3]	L. Hong, G. Chen, “Segment-Based Stereo Matching Using Graph Cuts,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004, Vol. 1, pp. 74-81.
[4]	J. Sun, N. Zang and H. Shum, “Stereo Matching Using Belief Propagation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, pp. 787-800, 2003.
[5]	J. Sun, Y. Li and S. B. Kang, “Symmetric Stereo Matching for Occlusion Handling,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005, Vol.2, pp. 399-406.
[6]	M. Bleyer and M. Gelautz, “A Layered Stereo Matching Algorithm Using Image Segmentation and Global Visibility Constraints,” International Society for Photogrammetry and Remote Sensing, Vol. 59, pp. 128-150, 2005.
[7]	N. Snavely, S. M. Seitz and R. Szeliski, “Photo Tourism: Exploring Photo Collections in 3D,” ACM Transaction on Graph, Vol. 25, pp. 835-846, 2006.
[8]	M. Gerrits and P. Bekaert, “Local Stereo Matching with Segmentation-Based Outlier Rejection,” The Third Canadian conference on Computer and Robot Vision, 2006, pp. 66-66.
[9]	S. T. Birchfield, B. Natarajan and C. Tomasi, “Correspondence as Energy-Based Segmentation,” Image and Vision Computing, Vol. 25, Issue. 8, pp. 1329-1340, 2007.
[10]	Z. F. Wang and Z. G. Zheng, “A Region Based Stereo Matching Algorithm Using Cooperative Optimization,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Anchorage, Alaska, USA, 2008, pp. 1-8.
[11]	G. Zhang, J. Jia, T. T. Wong and H. Bao, “Recovering Consistent Video Depth Maps via Bundle Optimization,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Anchorage, Alaska, USA, 2008, pp. 1-8.
[12]	D. Hoiem, A. A. Efros and M. Hebert, “Automatic Photo Pop-Up,” ACM Transaction on Graph, Vol. 24, pp. 577-584, 2005.
[13]	A. Saxena, J. Schulte and A. Y. Ng, “Depth Estimation using Monocular and Stereo Cues,” International Joint Conference on Artificial Intelligence, Hyderabad, India, 2007, pp. 2197-2203.
[14]	A. Saxena, M. Sun and A. Y. Ng, “Make3D: Depth Perception from a Single Still Image,” Twenty-Third AAAI Conference on Artificial Intelligence, Chicago, Illinois, USA, 2008, pp. 1571-1576.
[15]	Z. Zhang, “Flexible Camera Calibration By Viewing a Plane From Unknown Orientations,” International Conference on Computer Vision, 1999, pp. 666-673.
[16]	A. Criminisi, I. Reid and A. Zisserman, “Single View Metrology,” International Conference on Computer Vision, 1999, pp. 434-441.
[17]	A. Criminisi, I. Reid and A. Zisserman, “Single View Metrology,” International Journal of Computer Vision, Vol. 40, pp. 123-148, 2000.
[18]	A. Criminisi, “Single-View Metrology: Algorithms and Applications,” The German Association for Pattern Recognition,” 2002, pp. 224-239.
[19]	G. Wang, Z. Hu, F. Wu and H. Tsui, “Single View Metrology From Scene Constraints,” Image Vision Computing, Vol. 23, pp. 831-840, 2005.
[20]	H. Foroosh, X. Cao and M. Balci, “Metrology In Uncalibrated Images Given One Vanishing Point,” International Conference on Image Processing, 2005, pp. 361-364.
[21]	M. Shi and J. Y. Zheng, “A Slit Scanning Depth of Route Panorama from Stationary Blur,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005, Vol.1, pp. 1047-1054.
[22]	K. C. Zheng, S. B. Kang, M. F. Cohen and R. Szeliski, “Layered Depth Panoramas,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2007, pp. 18-23.
[23]	H. Nagahara, A. Ichikawa and M. Yachida, “Depth Estimation from the Color Drift of a Route Panorama,” International Conference on Intelligent Robots and Systems, 2008, pp. 4066-4071.
[24]	A. Rav-Acha, G. Engal and S. Peleg, “Minimal Aspect Distortion (MAD) Mosaicing of Long Scenes,” International Journal of Computer Vision, Vol. 78, pp. 187-206, 2008.
[25]	H. Bay, T. Tuytelarrs, L. J. V. Gool, “SURF: Speeded Up Robust Features,” European Conference on Computer Vision, 2006, Vol. 3951, pp. 404-417.
[26]	H. Bay, A. Ess, T. Tuytelarrs, L. J. V. Gool, “Speeded-Up Robust Features(SURF),” Computer Vision and Image Understanding, Vol. 110, pp. 346-359, 2008.
[27]	W. J. Tsai, T. K. Shih, Y. Y. Chao, H. Y. Zhong, “Luminance Adjustment in Image and Video,” Pacific-Rim Conference on Multimedia, Tainan, Taiwan, 2008, pp. 859-862.
[28]	J. Jia, J. Sun, C. Tang and H. Shum, “Drag-and-Drop Pasting,” ACM Transaction on Graph , Vol.25, pp. 631-637, 2006.
[29]	P. Perez, M. Gangnet, and A. Blake, “Poisson image editing,” ACM Transaction on Graph, Vol. 22, pp. 313–318, 2003.