跳到主要內容

臺灣博碩士論文加值系統

(44.210.83.132) 您好!臺灣時間:2024/05/27 02:37
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:周冠瑜
研究生(外文):CHOU-KUAN-YU
論文名稱:以深度學習應用在2D轉3D影像中結合 加權區域梯度的破洞填補法
論文名稱(外文):Deep Learning Applied with Weighted Local Gradients Method in Hole Filling for 2D to 3D Image Conversion
指導教授:楊權輝楊權輝引用關係
指導教授(外文):YANG,CHYUAN-HUEI
口試委員:蔡耀弘陳建彰楊權輝
口試委員(外文):TSAI, YAO-HONGChien-Chang ChenYANG,CHYUAN-HUEI
口試日期:2019-07-08
學位類別:碩士
校院名稱:玄奘大學
系所名稱:資訊管理學系碩士班
學門:電算機學門
學類:電算機一般學類
論文種類:學術論文
論文出版年:2019
畢業學年度:107
語文別:中文
論文頁數:45
中文關鍵詞:深度圖像生成演算法破洞填補深度學習型態學運算
外文關鍵詞:depth image based renderinghole fillinggradient vector flowdeep learningmorphological operations
Facebook:Guan Yu Zhou
相關次數:
  • 被引用被引用:0
  • 點閱點閱:388
  • 評分評分:
  • 下載下載:25
  • 收藏至我的研究室書目清單書目收藏:0
立體影像是透過2D轉3D影像技術所產生,深度影像繪圖法(Depth Image Based Rendering , DIBR)是將二維影像透過深度影像合成繪製多視角虛擬視圖。從基本DIBR系統繪製視圖導致影像位移失真、破洞等問題。對此本文從基本DIBR作修改,提出本文的方法以加權區域梯度向量流填補破洞,最初使用雙邊濾波器對DIBR之深度圖像進行前處理,作用於深度變化與相鄰的顏色,並消除雜訊對影像之扭曲變形。快速找出遮蔽區域的邊緣,其次分別使用梯度向量流方向尋找大區域破洞邊緣的顏色資訊並標記位置,以及使用形態學運算修補先前深度圖像上無法填補的小縫隙、破點,最後以大破洞的左側或右側加權區域梯度向量來填補該破洞。深度學習則為最近正廣泛使用在影像處理上,本研究利用深度學習找出最佳填補並結合加權區域梯度向量值。實驗結果顯示,本文提出的方法繪製出的視圖擁有更良好的品量。
關鍵詞:深度圖像生成演算法、破洞填補、梯度向量流、深度學習,型態學運算

The stereoscopic image generation from 2D to 3D conversion, Depth Image Based Rendering (DIBR) technique is 2D view to multi-angle virtual view by the single depth image. Traditional smoothing depth DIBR system generates geometric distortion in synthesis virtual view. In this research, a weighted local gradient guides DIBR with morphological operation hole-filling method is proposed. Applying the smooth filter in the DIBR firstly does the preprocessing of the depth image, which sharpens the discontinuous depth changes and smooth the neighboring depth of similar color and also keep noises from appearing on the warped images. To use the gradient vector flow to find the occlusion regions then used the morphological operations on the depth image with the background depth levels to keep the depth structure. We use the deep learning to find the best filling and use weighted local gradients from left or right side to render the large holes. Experimental results show that proposed method has good synthesis virtual view quality.

Keywords: depth image based rendering, hole filling, gradient vector flow, deep learning, morphological operations

摘要 I
Abstract II
致謝 III
目錄 IV
圖目錄 VII
第一章 序論 1
1.1. 研究背景與動機 1
1.2. 研究目的 2
1.3. 論文架構 3
第二章 文獻回顧 4
2.1 深度影像繪圖法 (Depth Image Based Rendering) 4
2.2 深度的梯度或梯度向量流的方法 10
2.3 補洞方法 12
2.4 深度學習 13
第三章 本文提出的方法 15
3.1 本文所利用的方法 15
3.1.1 深度影像繪圖法(Depth Image Based Rendering) 15
3.1.2 邊緣偵測(Edge Detection) 16
3.1.3 梯度向量流(Gradient Vector Flow) 16
3.1.4 中值濾波器(Median Filter) 17
3.1.5 雙邊濾波器(Bilateral Filter) 17
3.1.6 數學形態學(Mathematical Morphology) 18
3.1.7 主成分分析(Principal Components Analysis) 20
3.1.8 K-平均值分群法( (K-means Clustering) 20
3.1.9 深度學習(Deep Learning with Dropout) 21
3.2 本文用的步驟 22
第四章 本實驗結果 24
4.1 前處理 24
4.2 標記範圍 26
4.2.1 原圖標記 26
4.2.2 前處理標記 27
4.3 K-means 28
4.3.1 左視圖 28
4.3.2 左深度圖 28
4.3.3 右視圖 29
4.3.4 右深度圖 29
4.4 最後結果 30
4.4.1 左視圖 30
4.4.2 左深度圖 31
4.4.3 右視圖 31
4.4.4 右深度圖 32
第五章 結論與未來工作 33
參考文獻 34

[1]楊權輝、林恩瑞、陳柏翰,《以梯度向量流填補立體影像生成所產生破洞的方法》。2017 第七屆全國資訊管理前瞻技術研討會 ,台灣新竹。
[2]C. Cheng, C. Li and L. Chen, "A novel 2Dd-to-3D conversion system using edge information", IEEE Transactions on Consumer Electronics, vol. 56, no. 3, pp. 1739-1745, 2010.
[3]J. Cho, W. Song, H. Choi and T. Kim, "Hole Filling Method for Depth Image Based Rendering Based on Boundary Decision", IEEE Signal Processing Letters, vol. 24, no. 3, pp. 329-333, 2017.
[4]J. Choi, Y. Choe and Y. Kim, "Sparsity based depth estimation and hole-filling algorithm for 2D to 3D video conversion", Signals and Electronic Systems (ICSES), 2012 International Conference on, 2012.
[5]X. Deng, Q. Dai and X. Cao, "Depth and Residual Images Based Rendering", Chinese Journal of Electronics, vol. 25, no. 1, pp. 131-138, 2016.
[6]D. Doria and R. Radke, "Filling large holes in LiDAR data by inpainting depth gradients", Computer Vision and Pattern Recognition Workshops (CVPRW), 2012 IEEE Computer Society Conference on, pp. 65 - 72, 2012.
[7]M. Erofeev and D. Vatolin, "Multilayer semitransparent-edge processing for depth-image-based rendering", 3D Imaging (IC3D), 2016 International Conference on, 2016.
[8]S. Huang, J. Sun, Y. Yang, Y. Fang and P. Lin, "Multi-Frame Super-Resolution Reconstruction Based on Gradient Vector Flow Hybrid Field", IEEE Access, vol. 5, pp. 21669-21683, 2017.
[9]A. Jayachandran and V. Preetha, "Application of exemplar-based inpainting in depth image based rendering", 2016 IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), pp.1117 - 1121,2016.
[10]J. Jin, A. Wang and Y. Zhao, "A fast region-level 3D-warping method for depth-image-based rendering", Multimedia Signal Processing (MMSP), 2015 IEEE 17th International Workshop on, 2015.
[11]H. Kim and Y. Ro, "Multiview Stereoscopic Video Hole Filling Considering Spatiotemporal Consistency and Binocular Symmetry for Synthesized 3D Video", IEEE Transactions on Circuits and Systems for Video Technology, vol. 27, no. 7, pp. 1435-1449, 2017.
[12]Y. Lai, X. Lan, Y. Liu and N. Zheng, "An efficient depth image-based rendering with depth reliability maps for view synthesis", Journal of Visual Communication and Image Representation, vol. 41, pp. 176-184, 2016.
[13]W. Lie, C. Hsieh and G. Lin, "Key-Frame-Based Background Sprite Generation for Hole Filling In Depth Image-Based Rendering", IEEE Transactions on Multimedia, pp. 1-1, 2017.
[14]S. Lin, T. Shih and H. Hsu, "Filling holes in 3D scanned model base on 2D image inpainting", Ubi-media Computing and Workshops (Ubi-Media), 2017 10th International Conference on, pp. 101 - 106,2017.
[15]G. Luo and Y. Zhu, "Foreground Removal Approach for Hole Filling in 3D Video and FVV Synthesis", IEEE Transactions on Circuits and Systems for Video Technology, vol. 27, no. 10, pp. 2118-2131, 2017.
[16]K. Oh, S. Yea and Y. Ho, "Hole filling method using depth based in-painting for view synthesis in free viewpoint television and 3-D video", 2009 Picture Coding Symposium, pp. 233-236,2009.
[17]A. Oliveira, G. Fickel, M. Walter and C. Jung, "Selective hole-filling for depth-image based rendering", 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1886-1990, 2015.
[18]T. Phan, P. Shivakumara, B. Su and C. Tan, "A Gradient Vector Flow-Based Method for Video Character Segmentation", 2011 International Conference on Document Analysis and Recognition, pp. 1024 - 1028,2011.
[19]L. Po, S. Zhang and X. Xu, "A new multidirectional extrapolation hole-filling method for Depth-Image-Based Rendering", Image Processing (ICIP), 2011 18th IEEE International Conference on, pp. 2589 - 2592,2011.
[20]W. Qiongfei, Z. Yong and Z. Zhiqiang, "Infrared Image Segmentation Based on Gradient Vector Flow Model", 2015 Sixth International Conference on Intelligent Systems Design and Engineering Applications (ISDEA), pp. 460 - 462,2015.
[21]J. Seiler, S. Scholl, W. Schnurrer and A. Kaup, "Optimized processing order for 3D hole filling in video sequences using frequency selective extrapolation", 2016 Picture Coding Symposium (PCS), 2016.
[22]M. Solh and G. AlRegib, "Hierarchical Hole-Filling(HHF): Depth image based rendering without depth map filtering for 3D-TV", 2010 IEEE International Workshop on Multimedia Signal Processing, pp. 87 - 92,2010.
[23]W. Sun, O. Au, L. Xu, Y. Li and W. Hu, "Novel temporal domain hole filling based on background modeling for view synthesis", 2012 19th IEEE International Conference on Image Processing, pp. 2721 - 2724,2012.
[24]C. Yang, C. Wu and P. Chen, "A HOLE FILLING METHOD FOR GENERATING STEREOSCOPIC IMAGE BY EDGE ORIENTED WITH MORPHOLOGICAL OPERATIONS", International Conference on INTERNET STUDIES, Bali, Indonesia., 2017.
[25]H. Yin, L. Zhang, G. Luo and Y. Zhu, "A new disocclusion filling approach in depth image based rendering for stereoscopic imaging", 2015 International Conference on Control, Automation and Information Sciences (ICCAIS),pp. 313 - 317, 2015.
[26]H. Zhou, X. Li, G. Schaefer, M. Celebi and P. Miller, "Mean shift based gradient vector flow for image segmentation", Computer Vision and Image Understanding, vol. 117, no. 9, pp. 1004-1016, 2013.
[27]L. Zhu, M. Yu, G. Jiang, X. Mao, S. Fu and T. Luo, "A New Virtual View Rendering Method based on Depth Map for 3DTV", Procedia Engineering, vol. 15, pp. 1115-1119, 2011.
[28]X.Gao, J.Zhang, Z.Wei,“Deep Learning for Sequence Pattern Recognition”, 2018 IEEE 15th International Conference on Networking, Sensing and Control (ICNSC), Zhuhai, China, 27-29 March 2018, pp.1-6
[29]P.Kinghorn, L.Zhang, L.Shao“Deep Learning based Image Description Generation ”, 2017 International Joint Conference on Neural Networks (IJCNN) , Anchorage, AK, USA, 14-19 May 2017, pp.919-926
[30]L.Merrick, Q.Gu,“Exploring the Use of Adaptive Gradient Methods in Effective Deep Learning Systems”, 2018 Systems and Information Engineering Design Symposium (SIEDS), Charlottesville, VA, USA, 27-27 April 2018, pp.220-224
[31]H.Petersson, D.Gustafsson D.Bergstr¨om,“Hyperspectral Image Analysis using Deep Learning - a Review”, 2016 Sixth International Conference on Image Processing Theory, Tools and Applications (IPTA), Oulu, Finland, 12-15 Dec. 2016, pp.1-6
[32]Q.Wu, Y.Liu, Q.Li, S.Jin ,F.Li,“The Application of Deep Learning in Computer Vision”, 2018 6th Iranian Joint Congress on Fuzzy and Intelligent Systems (CFIS), Kerman, Iran, 28 Feb.-2 March 2018, pp.75-78
[33]U.Yadav, D.K.Xaxa, S.Verma, C.Mahobiya, “A deep learning based character recognition system from multimedia document ”,2017 Innovations in Power and Advanced Computing Technologies (i-PACT) , Vellore, India, 21-22 April 2017
[34]Wiki- Principal components analysis, https://zh.wikipedia.org/wiki/%E4%B8%BB%E6%88%90%E5%88%86%E5%88%86%E6%9E%90
[35]Dropout,https://medium.com/%E6%89%8B%E5%AF%AB%E7%AD%86%E8%A8%98/%E4%BD%BF%E7%94%A8-tensorflow-%E4%BA%86%E8%A7%A3-dropout-bf64a6785431
[36]Wiki- k-means clustering, https://zh.wikipedia.org/wiki/K-%E5%B9%B3%E5%9D%87%E7%AE%97%E6%B3%95


QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top