臺灣博碩士論文加值系統

為提供三維立體顯示器之多視點影像來源，傳統的多次渲染法效率相當低落，而目前最常見的多視點影像生成法稱為基於深度影像繪圖法DIBR(Depth-Image-Based Rendering)，此演算法之效能與視點數無關，但會產生一些畫面上的瑕疵，這些瑕疵乃源自於單一視點畫面所提供的資訊限制。本論文結合電腦圖學特性與DIBR演算法，首先利用深度資訊與視點位移量預先計算出破洞將會產生之區域，並以此作為遮罩進行部份場景重繪，藉此得到單一視點畫面所缺失的額外資訊，以消除其畫面瑕疵，由於僅針對破洞區域進行重繪，相較於其它額外資訊修補法，可以減少不必要的影像資訊，降低多餘影像產生的額外資料量。實驗結果顯示，我們所需之額外資料量為多次渲染法的0.26%，雙眼立體影像轉換多視點影像法的0.64%及基於補償性深度繪圖法的39%。

To provide multi-view image for three-dimensional display, the efficiency of traditional rendering multiple times method is very low. There is an alternative called depth-image-based rendering (DIBR), which is a view number independent algorithm. However, this algorithm produces some visible artifacts due to the single image does not contain enough information. Hence we proposed an algorithm that combines the technology of computer graphic and DIBR. By using the viewpoint moved distance and the depth information, we predict the hole area, then take it as a mask for hole area partial re-rendering and supply those information to reduce those artifacts. The additional information of our algorithm for hole-filling is only the hole area image, therefore we can reduce the additional data. According to experimental result, the data volume for the proposed method is only 0.26% that of the rendering multiple times method, 0.64% that of the stereo to multi-view method and 39% that of the compensated DIBR method.

論文審定書 i
摘 要 ii
Abstract iii
目錄 iv
圖目錄 vi
表目錄 viii
第1章 簡介 1
1.1 論文概述 1
1.2 論文貢獻 2
1.3 論文架構 2
第2章 背景知識及相關文獻 3
2.1 視覺與深度感知系統 3
2.1.1 單眼視覺(Monocular Vision) 3
2.1.2 雙眼視覺(Binocular Vision) 4
2.2 3D立體顯示技術 5
2.2.1 眼鏡式立體顯示器 5
2.2.2 裸眼式立體顯示器 6
2.2.3 雙眼視差及深度感知 7
2.3 多視點影像來源 8
第3章 基於深度影像繪圖法(Depth Image Based Rendering) 9
3.1 影像位移 10
3.1.1 間距縮小及遮擋(Minification and Occlusion) 11
3.1.2 間距擴張(Magnification) 12
3.2 破洞填補 16
3.2.1 前置處理法(Pre-processing Approach) 16
3.2.2 後置處理法(Post-processing Approach) 17
3.2.3 額外資訊處理法(Additional Information Approach) 17
3.3 DIBR分析 18
第4章 研究方法 19
4.1 深度處理(Depth Processing) 20
4.1.1 Z值轉換(Z-value Transformation) 21
4.1.2 位移基底計算(Shifting Base Calculation) 22
4.1.3 破洞區域計算(Hole Area Calculation) 23
4.2 部份渲染(Partial Rendering) 24
4.3 基於深度影像繪圖法(DIBR) 25
第5章 實做方法及研究結果 26
5.1 深度處理 26
5.2 部份渲染 28
5.3 DIBR 29
5.4 額外資料量比較 31
5.5 生成畫面評估 32
5.6 總結 34
第6章 結論及未來目標 37
參考文獻 38

[1]Kinki University Faculty of Medicine Department of Ophthalmology, Sep, 2016, http://www.med.kindai.ac.jp/optho/english/olaboratory.html
[2]C. E. West and D. G. Hunter, “Displacement of optical centers in over-the-counter readers: a potential cause of diplopia,” Journal of American Association for Pediatric Ophthalmology and Strabismus (JAAPOS), vol. 18, is. 3, pp. 293-294, 2014.
[3]Radboud University Faculty of Science Virtual Classroom Biology, Sep, 2016, http://www.vcbio.science.ru.nl/en/image-gallery/stereo/
[4]Autostereoscopic Displays, Sep, 2016, http://www.3d-forums.com/threads/autostereoscopic-displays.1/
[5]J. K. Cho, “Multi-view image generation using compensated depth-image-based rendering for 3d displays,” Master’s Thesis, Institute of Computer & Communication, National Cheng Kung University, 2007.
[6]L. McMillan, “An image-based approach to three-dimensional computer graphics,” PhD thesis, Dept. of Computer Science, University of North Carolina, Chapel Hill, 1997.
[7]J. Compen, “3D graphics rendering for multiview displays: Using programmable shaders on graphics cards,” Koninklijke Philips Electronics N.V. 2005, Technical note TN-2004/00920, 2005.
[8]C. Fehn, “Depth-image-based rendering (DIBR), compression, and transmission for a new approach on 3D-TV,” SPIE Stereoscopic Displays and Virtual Reality Systems XI, vol. 5291, pp. 93-104, 2004.
[9]L. Zhang and W. Tam, “Stereoscopic image generation based on depth images for 3D TV,” IEEE Transactions on Broadcasting, vol. 51, iss. 2, pp.191-199, 2005
[10]W. Y. Chen, Y. L. Chang and S. F. Lin, “Efficient depth image based rendering with edge dependent depth filter and interpolation,” IEEE ICME, pp. 1314-1317, 2005.
[11]Y. R. Horng, Y. C. Tseng, and T. S. Chang, "Stereoscopic images generation with directional Gaussian filter,” IEEE ISCAS, pp. 2650-2653, 2010.
[12]C. Vázques, W. J. Tam, and F. Speranza, “Stereoscopic imaging: filling disoccluded areas in depth image-based rendering,” SPIE Three-Dimensional TV, Video, and Display V, vol. 6392, pp. 63920D, 2006.
[13]L.M. Po, S. Zhang, X. Xu, and Y. Zhu, “A new multidirectional extrapolation hole-filling method for depth-image-based rendering,” IEEE ICIP, pp. 2637-2640, 2011.
[14]Z. Tauber, Z. N. Li, and M. S. Drew, “Review and preview: disocclusion by inpainting for image-based rendering,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 37, iss. 4, pp. 527-540, 2007.
[15]K. Luo, D. Li, Y. Feng, and M. Zhang, “Depth-aided inpainting for disocclusion restoration of multi-view images using depth-image-based rendering,” Journal of Zhejiang University SCIENCE A, vol. 10, pp. 1738-1749, 2009.
[16]I. Daribo and B. P. Popescu, “Depth-aided image inpainting for Novel View Synthesis,” IEEE MMSP, pp. 167-170, 2010.
[17]X. Xu1, L. M. Po, C. H. Cheung , L. Feng, K. H. Ng, and K. W. Cheung, “Depth-aided exemplar-based hole filling for DIBR view synthesis,” IEEE ISCAS, pp. 2840-2843, 2013.
[18]J. Shade, S. Gortler, L. W. He, and R. Szeliski, “Layered depth images,” ACM SIGGRAPH, pp. 231-242, 1998.
[19]MSU Stereo To Multiview 3D Video Conversion - content creation technology for glasses-free 3D displays, Sep, 2016, http://www.compression.ru/video/3d_display_video/index_en.html
[20]C. Zhu and S. Li, “Multiple reference views for hole reduction in DIBR view synthesis,” IEEE BMSB, pp. 1-5, 2014.
[21]Y. C. Wu, “Image synthesis technology for multi-view 3d displays,” Master’s Thesis, Institute of Computer & Communication, National Cheng Kung University, 2008