跳到主要內容

臺灣博碩士論文加值系統

(18.97.9.172) 您好!臺灣時間:2025/02/16 21:10
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:林宏達
研究生(外文):Hong-Da Lin
論文名稱:利用紋理合成的快速影像補繪法則之研究
論文名稱(外文):A FAST ALGORITHM FOR IMAGE INPAINTING VIA TEXTURE SYNTHESIS
指導教授:鄭嘉慶鄭嘉慶引用關係
指導教授(外文):Jia-Ching Cheng
學位類別:碩士
校院名稱:大同大學
系所名稱:通訊工程研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2006
畢業學年度:94
語文別:英文
論文頁數:46
中文關鍵詞:影像補繪紋理合成
外文關鍵詞:image inapintingtexture synthesis
相關次數:
  • 被引用被引用:0
  • 點閱點閱:189
  • 評分評分:
  • 下載下載:16
  • 收藏至我的研究室書目清單書目收藏:0
在此論文中,我們提出一個快速紋理合成的補繪法則,此方法是一種基於像素的紋理合成方法。為來填補未知區域中的像素,基於像素的紋理合成方法需要一個正方形的比對視窗,而此視窗中已知的像素分佈成固定方向的L型區域。這種方式在填補矩形區域時,因為是水平的搜尋方向所以能有較好的效率。然而,當填補不規則未知區域時,由於搜尋方向不是固定的,因此降低填補的效率。
在這裡我們的目標是提出一個有效率的填補不規則區域的補繪法則,其關鍵在於建構一個可以改變搜尋方向的比對視窗。我們提出一個將二維的比對視窗轉換成一維的環形向量的鋸齒狀對應方式(zigzag mapping)。我們的方法設定未知區域輪廓線通過搜尋視窗的中心點,並將視窗分成相等的兩區域,即已知像素的區域和未知像素的區域。此鋸齒狀的對應允許已知區域中的像素所組成的一維片段在環形向量中是一連續的片段。如此,不只提高搜尋的準確性,也加快了補繪的速度。
我們成功的利用數張影像來展示我們提出的補繪法則,實驗顯示,在搜尋效率和準確度上都能得到令人滿意的結果。
In this thesis, we present a fast algorithm for image inpainting via texture synthesis. This method is categorized as a pixel-based texture synthesis technique. To assign the unknown target pixel in area yet to be filled in, the pixel-based texture synthesis technique requires a matching template which is a square matching window consisting of known pixels in an L-shape region of fixed direction. This technique can be applied efficiently to fill in regions of rectangle shape due to their horizontal searching direction; however, it does not function efficiently to fill in regions of irregular shape which have searching direction of arbitrary orientation.
The method we proposed aims at filling in regions of irregular shape with fair efficiency. The key is to construct a matching window of searching direction with variable orientation. We devise a zigzag mapping method which converts the two-dimensional matching window into a cyclic vector. The contour line of unknown region is designated to pass the center of the matching window and separate it into two equal halves, region of known pixels and region of unknown pixels, respectively. Our zigzag pattern allows the pixels of the known region to form a continuous segment in the cyclic vector. By doing so, we have raised not only searching accuracy but also searching speed.
We successfully demonstrate the proposed inapinting algorithm by using several images and obtain promising experimental results in terms of both efficiency and effectiveness.
ACKNOWLEDGEMENT…………………………………………………………I
ENGLISH ABSTRACT………………………………………………………II
CHINESE ABSTRACT………………………………………………………III
TABLE OF CONTENTS………………………………………………………IV
LIST OF FIGURES…………………………………………………………VI
LIST OF TABLES…………………………………………………………VIII
CHAPTER
1 INTRODUCTION …………………………………………………1
1.1 Overview………………………………………………………………1
1.2 The Proposed Scheme……………………………………2
1.3 Thesis Outline……………………………………………3
2 OBJECT EXTRACTION……………………………………………………5
2.1 Introduction ………………………………………………………...5
2.2 Object Extraction Methods…………………………………………5
2.2.1 Manual selection…………………………………………..6
2.2.2 Intelligent scissors ……………………………………...7
2.2.3 Motion-based segmentation …………………………….8
2.2.4 Text Extraction ……………………………………………9
2.3 Summary…………………………………………………………..10
3 IMAGE TEXTURE SYNTHESIS …………………………………….11
3.1 Introduction ……………………………………………….11
3.1.1 What is a Texture? ……………………………………..11
3.1.2 What is Texture Synthesis? ………………...……………13
3.2 Wei and Levoy Texture Synthesis ………………………………..14
3.2.1 Overview………………………………………………...14
3.2.2 Neighborhood……………………………………………16
3.2.3 Algorithm ……………………………………………….18
4 INPAINTING ALGORITHM ………………………………………...22
4.1 Introduction ………………………………………………………22
4.2 The Algorithm…………………………………………………….22
4.2.1 Search window ………………………………………….22
4.2.2 Inpainting ………………………………………………..26
4.2.3 Filling priority ………………………………………...30
4.3 Overview………………………………………………………… 33
4.4 Summary …………………………………………………………36
5 EXPERIMENTAL RESULTS ………………………………………...37
5.1 Introduction………………………………………………………..37
5.2 Test images ……………………………………………………….37
5.2.1 Door image ………………………………………………39
5.2.2 R23 texture image………………………………………..40
5.2.3 Desert image ……………………………………………..41
5.2.4 Grassland image …………………………………………42
5.3 Summary…………………………………………………………. 43
6 CONCLUSIONS………………………………………………………44
REFERENCES …………………………………………………………...45
[1]A. A. Efros and T. K. Leung, “Texture synthesis by non-parametric sampling,” Proceedings of International Conference on Computer Vision, pp.1033-1038, Sep. 1999.
[2] A. C. Kokaram, R. D. Morris, W. J. Fitzgerald, and P. J. W. Rayner, “Detection of missing data in image sequences,” IEEE Transactions on Image Processing, vol. 4, p.p.1496-1508, Nov. 1995.
[3] A. C. Kokaram, R. D. Morris, W. J. Fitzgerald, and P. J. W. Rayner, “Interpolation of missing data in image sequences,” IEEE Transactions on Image Processing, vol. 4, p.p.1509-1519, Nov. 1995.
[4]P. F. Felzenszwalb, and D. P. Huttenlocher, “Image segmentation using local variation,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.98 –104, June 1998.
[5]L. Y. Wei and M. Levoy, “Fast texture synthesis using tree-structured vector quantization,” Proceedings of SIGGRAPH, pp.479-488, July 2000.
[6]A. Criminisi, P. Perez, K. Toyama, “Region filling and object removal by exemplar-based image inpainting”, IEEE Transactions on Image Processing, vol. 13 no. 9, September 2004.
[7]J. F. Wang, H. J. Hsu, and H. M. Wang,” Constrained texture synthesis by scalable sub-path algorithm,” IEEE International Conference on Multimedia and Expo, vol. 1, pp.635-638, June, 2004.
[8]A. Nealen and M. Alexa, “Hybrid texture synthesis,” Proceedings of ACM International Conference, vol.44, pp.97-105, 2003.
[9]A. Criminisi, P. P�臆ez, and K. Toyama, “Object removal by exemplar-based inpainting,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol.2, pp.18-20, June 2003.
[10]M.Ashikhmin, “Synthesizing natural textures,” in Proceedings of ACM Symp. on Interactive 3D Graphics, pp.217-226, March 2001.
[11]H.Yamauchi, J. Haber and H. P. Seidel, “Image restoration using multiresolution texture synthesis and image inpainting,” Proceedings of Computer Graphics International, pp.120-125, July 2003.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top