Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/16567
標題: SIFT演算法應用於三維建物模型表面影像拼接之研究─以中興大學圖書館為例
The Study of Perspective Texture Mapping on 3D Building Models Based on SIFT Algorithm-The campus of National Chung-Hsing University Library as an Example
作者: 莊凱婷
Juang, Kai-Ting
關鍵字: Image Processing
影像處理
SIFT
RANSAC
feature matching
SIFT
RANSAC
特徵匹配
出版社: 土木工程學系所
引用: 1.黃文利,2001.06,"近景攝影測量應用於三維建物模型側面影像敷貼之研究",成功大學測量工程碩士論文 2.傅秉綱,2002.06,"三維建物模型表面影像敷貼自動化之研究",成功大學測量工程碩士論文 3.林后駿,2005.06,"三維房屋模型實景紋理影像製作與敷貼之研究",中央大學土木工程碩士論文 4.孫敏、胡婷婷、朱強,2007.01,"面向城市3維建模的影像校正方法",測繪通報,第四期,p42-45 5.李寒、牛紀楨、郭禾,2007.05,"基於特徵點的全自動無縫圖像拼接方法",計算機工程與設計,28(9),p2083-2085 6.戚世貴、戚素娟,2008.01,"一種基於圖像特徵點的圖像匹配算法",中國科技核心期刊,第二七卷第一期,p3-4、15 7.王聖鐸、曾義星、傅秉綱,2008.06,"以真實影像自動化敷貼三維建物模型表面之研究",航測及遙測學刊,第十三卷第二期,p74-83 8.趙敬紅、孫敏、向南平,2008.06,"利用對應矩陣進行影像校正的誤差分析",測繪通報,2008(11),p18-20 9.何明哲、王崴、韓華亭,2009.03,"基於SIFT特徵的地貌圖像匹配研究",計算機應用與軟件,26(3),p228-230 10.吳究、 張奇、劉銘哲,2009.06,"基於尺度不變特徵點轉換之精密多元影像套合",航測及遙測學刊,14(2),p141-155 11.黃漢哲,2009.08,"SIFT 演算法應用於航測影像拼接之研究",中山大學海洋環境及工程學系碩士論文 12.閔曉媛、閆曉光,2010.01,"基於SIFT 演算法的自動全景圖生成演算法",科技創新導報,29,p97 13.嚴磊、汪增福,2010.1,"基於多局部特徵匹配的全自動圖像拼接",計算機應用與軟件,27(10),p5-7 14.韋虎、張麗豔、劉勝蘭、石春琴,2010.04,"基於SIFT 圖像特徵匹配的多視角深度圖配准演算法",計算機輔助設計與圖形學學報,22(4),p654-661 15.陳家琪、郝妍,2010.07,"一種改進的SIFT 特徵反向匹配演算法",軟體時空,27(3),p209-211 16.孫豔麗、李建海、王玲玲、孫晶,2010.07,"基於SIFT 的多焦距圖像特徵點提取演算法",自動化技術,23,p116-118 17.鄒北驥、阮鵬、向遙、賀加貝,2010.08,"一種精確匹配的全景圖自動拼接演算法",計算機工程與科學,32(8),p60-63 18.胡影峰,2011.01,"基於SIFT特徵匹配運算元的三維重建方法研究",PROCESS AUTOMATION IN STRUMENTATION,32(1),p30-32 19.夏春林、王佳奇,2011.01,"3DGIS中建築物三維建模技術綜述",測繪科學,36(1),p70-72 20.曹楠、王萍,2011.02,"基於SIFT 特徵匹配的圖像無縫拼接算法",計算機與應用化學,28(2),p242-244 21.康志忠,2008.12,"近景數碼影像中牆面紋理自動拼接方法的研究",中國圖像圖形學報,12(12),p2368-2373 22.陳正軒,2006.06,"以視訊影像進行三維房屋模型實景紋理敷貼之研究",中央大學土木工程碩士論文 23.Martin A. Fischler and Robert C. Bolles,(1981),"Random Sample Consensus: A Paradigm for Model Fitting with Apphcatlons to Image Analysis and Automated Cartography",Graphics and Image Processing,24(6),pp.381-395 24.David G. Lowe,(2004), Distinctive image features from 116 scale-invariant Key points. International Journal of Computer Vision, No.60, pp. 91-110 25.David G. Lowe,(2006), Automatic Panoramic Image Stitching sing Invariant Features. International Journal of Computer Vision, No.74(1), pp.59–73 26.Lindeberg T.(1998), Feature detection with automatic scale selection.117 International Journal of Computer Vision, No.30(2),pp.79-116
摘要: 虛擬城市(Virtual City)的真實性隨著貼圖技術的發展而逐漸進步,尤其在現今使用真實影像作為三維物體表面的敷貼材質越來越廣泛,也使得在展現三維場景或城市模型成果時越來越能表達真實的感覺。然因建立三維建物敷貼材質時,常因建物表面積過大,或者是建物與其它物體之間的間距太小,導致拍攝時單幅影像包含紋理不完全,使每個模型面的材質可能需要由多張像片拼接而成,使得如何得到一個完整的模型面材質影像便成為一個重要的研究課題。 本研究利用SIFT(Scale Invariant Feature Transform)演算法進行多幅近景影像拼接的工作,SIFT演算法為利用電腦視覺進行特徵點萃取的演算法,其特徵點具有抗尺寸、角度及亮度改變的特性,且其特徵點具獨特性、多量性及均勻性的優點,搭配RANSAC(RANdom SAmple Consensus)除錯機制挑選出正確且具代表性的特徵點後,使重疊區域的影像拼接成一張完整且連續的模型面材質,供後續建物模型敷面的擷取。 本研究選定中興大學圖書館做為研究對象,以非量測型數位相機拍攝建物表面影像後,使用CS5商業軟體求解校正的建物影像,接著使用SIFT演算法進行擷取面的影像拼接,去除拼接影像遮蔽情形及微調整欲擷取面後,成功的產生了完整的紋理材質供三維模型使用,最後利用Sketchup pro及CAD平面圖建立比例尺1:1的三維模型,測試的模型敷貼成果獲得令人滿意的視覺效果,證實了本文所提方法之可行性。
With the progress of the texture-mapping techniques, the authenticity of the Virtual city has been radically improved. Nowadays, photo-realistic texture mapping of 3D objects is more diverse, and also presents more realistic quality of view in the 3D show or in the city model. However, when setting up 3D texture mapping material, the result of single image shooting is usually not complete because the size of buildings is too big or the space between the main building and other objects is too small. Hence, how to get a complete model of surface texture image has become an important research topic. In this study, SIFT (Scale Invariant Feature Transform) algorithm was used for image stitching multiple close-range work, and was used in the computer vision to perform feature extraction in good condition. The extracted SIFT features are invariant to image scale, rotation, noise and change in illumination Moreover, it is a robust and abundant feature extraction algorithm, and the feature points with the advantages of uniformity. Through additional RANSAC outlier removal procedure to select correct and characteristic feature points, the overlapping area of the image combines into a complete and continuous stitching model of surface material, which provides covered model buildings for subsequent capture. I selected National Chung Hsing University Library as a study example for shooting the building by non-measurement type digital camera. Next step is to correct image through commercial software CS5, and then use the SIFT algorithm for stitching images captured surface. After removing conditions and micro-mosaic image and adjusting to the prefer capture, the three-dimensional model was successfully produced for the texture material. Finally, using Sketch up pro and CAD to establish 1:1 scale 3D model layout got a perfect visual effects. Therefore, this study confirmed the feasibility of the proposed method.
URI: http://hdl.handle.net/11455/16567
其他識別: U0005-2607201120581000
文章連結: http://www.airitilibrary.com/Publication/alDetailedMesh1?DocID=U0005-2607201120581000
Appears in Collections:土木工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.