(3.236.231.61) 您好!臺灣時間:2021/05/15 23:43
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:簡國綱
研究生(外文):Kuo Kang Chien
論文名稱:3D掃瞄資料貼圖及顏色定位之研究
論文名稱(外文):Texture mapping and 3D registration using color information for multiple scan data
指導教授:姚宏宗姚宏宗引用關係
指導教授(外文):Hong-Tzong Yau
學位類別:碩士
校院名稱:國立中正大學
系所名稱:機械系
學門:工程學門
學類:機械工程學類
論文種類:學術論文
論文出版年:2001
畢業學年度:89
語文別:中文
論文頁數:97
中文關鍵詞:掃瞄資料貼圖顏色定位
外文關鍵詞:texturemappingscanregistration
相關次數:
  • 被引用被引用:4
  • 點閱點閱:273
  • 評分評分:
  • 下載下載:56
  • 收藏至我的研究室書目清單書目收藏:1
在現今三次元掃瞄機已可將物體的3D資料準確地掃出,但在物體上的顏色資訊卻往往闕如,針對於此,本文研究的一項重點,就是將掃瞄機所掃的3D資料與手持相機所得相片作結合,使準確的3D資料同時能有顏色的資訊。其中我們在此主題上有三重點: (1)相片與模型定位(2)多圖片顏色混成(3)網格簡化貼圖。在(1)中,我們以手動點選的方法,選擇至少6點對,再以相機定位(camera calibration)的方法,將相片與3D資料的轉換式算出,再依此將相片與3D資料加以結合。在(2)中,將模型貼圖時所建立的圓柱投影座標作為主要平面,再將多張相片轉換至此,利用金字塔法將平面上的顏色混成,使之顏色能平順、均勻,在以此平順的圓柱平面貼到模型上,就可達色彩均勻的模型。在(3)中, 首先將模型資料量簡化,用頂點幾合判斷、曲率大小及容許誤差來過濾掉三角網格,得到一個資料量小且在容許誤差下的模型,再配合上圓柱貼圖,將圖不變形地貼上,如此,我們將可得到一兼具外觀且易操作的模型。 藉由以上三步驟,我們可將相片與掃瞄資料作一顏色均勻的結合,並且可藉由貼圖及三角網格簡化的方式,達到資料少且彩現完整的模型。
在現今多筆掃瞄資料的定位,在許多的研究皆是以ICP(iterative closest point)來做定位,但ICP需有良好的初始位置,才能減少疊代次數並增加精準度。故此,本研究的另一重點就在於利用顏色資料作三角網格的初始定位。作法是將有顏色資訊的掃瞄資料上,藉由彩色圖片對應點的方法,將二筆資料的對應點找出,再由這些點對經由SVD(singular value decomposition)法將轉換矩陣求出,如此可快速將二筆資料作初始定位,以利於ICP做精確定位。
This paper describes how to combine the scan data with the pictures, which were taken from the camera. This will lead to a model, which has exactly position data and smooth color view. This model can be used in CAD/CAM/CAE with its high accuracy, and also can be showed in computer vision with its color texture mapping. For this purpose, there have three steps to do: (1) combine 2D picture and 3D scan data, (2)mosaic many view of picture to make the texture model smoothly, (3)reduce the model’s triangle number. In (1), select many pairs of point on the model and picture. Use these pairs of point to retrieve the camera matrix. Then, use this matrix to relate 3D and 2D data. In (2), project all pictures on the cylinder plane. Then, use 2D image mosaic technique to smooth the difference between the pictures. Here, pyramid method is used. In (3), reduce the data amount by triangular geometry、curvature、error tolerance. Then, map the texture on the model. Through these three steps, a model, which has smooth color view and exact position data was found. This model also can be easily operated in the computer vision.
Many researches use ICP (iterative closet point) to regitrate the 3D datas, but ICP must have a good initial position to reduce iterative times. Our research will use color information to registrate automatically 3D datas’s initial position. Take the color scan datas, which were taken from 3D color scanner or from previous method to do image process. In image process, the match point pairs between the pictures will be found. Then, use 2D-3D relation to project these 2D points to the 3D space. Finally, use these 3D point pairs to retrieve the rotation-translation matrix. This matrix can be used to registrate the datas to the good initial position.
一、緒論……………………………………………………….………..1
1.1前言……………………………………………………………….1
1.2研究動機及目的……………………………….………………...4
1.3文獻回顧…….…………………………………………………..6
1.3.1彩色圖片分析及對應……………………………………6
1.3.2空間定位.…………………………………………………7
1.3.3相機特性及相片顏色混成………………………………8
1.3.4物體位置對正…………………………………………….9
1.3.5由相片取得3D資料……………………………………..10
1.4 研究方法……………………………………………………….10
二、貼圖座標與點選貼圖……………….……………………..……...12
2.1圓柱座標…………………………………………………….12
2.2點選貼圖……………………………………………………….14
2.3 掃瞄線取色貼圖……………………………………………….16
2.4 實際範例……………………………………………………….20
三、多張相片貼圖與混成……………….……………………..……...27
3.1平面影像混成(Image Mosaic)……………………………28
3.1.1高斯金字塔(Gaussian Pyramid)……………29
3.1.2拉普拉斯金字塔(Laplacian Pyramid)...31
3.1.3金字塔運算與顏色混成(Image Mosaic)...32
3.2多張相片貼圖與混……………………………………..35
3.3實際範…………………………………………………..37
四、三角網格簡化與貼圖……………….……………………..……...41
4.1網格資料簡………………………………………………….41
4.1.1網格頂點幾何型態判斷……………………………….42
4.1.2頂點移除與網格重建…………………………………..44
4.1.3誤差計算………………………………………………..47
4.2網格簡化貼……………………………………………….51
4.3實例範例………………………………………………….53
五、在未定位相機之圖片中尋找對應特徵點……………….………57
5.1圖片特徵點尋找……………………………………………57
5.1.1高斯平滑化(Gaussian smooth)…………………………58
5.1.2彩色特徵點尋找(Harris detector)……………………60
5.2圖片間特徵點之對……………………………………..62
5.2.1局部常態化(local normalize)…………………………..62
5.2.2彩色特徵向量……………………………………………63
5.2.3對應力(strength of match)…………………………….66
5.2.4相對過濾法………………………………………………69
5.3實際範例………………………………………………..71
六、多筆資料之顏色定位………………………………………………81
6.1圖片與掃瞄資料之結合……………………………..…81
6.2對應點定…………………………………………………..82
6.3實際範例………………………………………………..85
七、結論…………………………………………………………………92
參考文獻………………………………………………………………..95
參考文獻
[1]P. Montesinos, V. Gouet, R. Deriche. Differential
Invariants for Color Images. IEEE Pattern Recognition,
Vol.1,1998 , pp. 838-840.
[2] V.Gouet, P. Montesinos, D, Pele. Stereo Matching of Color
Images Using Differential Invariants. IEEE Image
Processing, 1998, Vol.2, pp. 152-156.
[3] Z. Zhang, R. Deriche, O. Faugeras,Q. T. Luong. A Robust Technique for Matching TwoUncalibrated Images Through the Recovery of the Unknown Epipolar Geometry. Research report n° 2273 INRIA Sophia-Antipolis, France, 1994.
[4]G. Xu, Z. Zhang. Epipolar Geometry in Stereo, Motion and Object Recognition. Kluwer Academic Press, 1996.
[5]A. E. Johnson, S. B. Kang. Registration and Integration of Textured 3-D Data. IEEE 3-D Digital Imaging and Modeling, 1997, pp.234-241.
[6]S. Weik. Registration of 3-D Partial Surface Models Using Luminance and Depth Information. IEEE 3-D Digital Imaging and Modeling, 1997, pp.93-100.
[7]C. S. Cheng, Y. P. Hung, J. B. Cheng. A Fast Automatic Method for Registration of Partially-Overlapping Range Images. IEEE Computer Vision, 1998, pp.242-248.
[8] 陳俊諺. 利用3D多重掃瞄資料建構多面體價購之實體
模型. 國立中正大學機械工程研究所碩士論文, 2000.
[9]Y. Nomura, D. Zhang, Y. Sakaida, S. Fujii. 3-D Object Pose Estimation Based on Iterative Image Matching: Shading and Edge Data Fusion. IEEE Pattern Recognition, 1996, vol.1, pp.513-517.
[10]O. Faugeras. Three-Dimensional Computer Vision. The MIT Press 1993.
[11]P. J. Burt, E. H. Adelson. A Multiresolution Spline With Application to Image Mosaics. ACM Transactions on Graphics, October 1983, vol.2, No.4, pp.217-236
[12]C. T. Hsu, J. L. Wu. Multiresolution Mosaic. IEEE Transactions on Coinsumer Electronics, Noverber 1996, vol.42, No.4, pp.981-990
[13]R. Szeliski, H. Y. Shum. Creating Full View Panoramic Image Mosaics and Environment Maps. ACM International Conference on Computer Graphics and Interactive Techniques, August 1997.
[14]P. Bao, D. Xu. Panoramic Image Mosaics via Complex Wavelet Pyramid. IEEE Systems, Man, and Cybernetics, 1998, vol.5, pp.4614-4619.
[15]M. Kayanuma, M. Hagiwara. A New Methoc to Detect Object and Estimate the Position and the Orientation from an Image Using a 3-D Model Having Feature Points. IEEE
Systems, Man, and Cybernetics, 1999, vol.4, pp.931-936.
[16]P. Wunsch, G. Hirzinger. Registration of CAD-Models to Images by Interative Inverse Perspective Matching. IEEE Pattern Recognition, 1996, vol.1, pp.78-83.
[17]Q. Chen, H. Wu, T. Shioyama, T. Shimada. 3D Head Pose Estimation Using Color Information. IEEE Multimedia Computing and Systems, 1999, vol.1, pp.697-702.
[18]Q. Chen, H. Wu, T. Fukumoto, M. Yachida. 3D Head Pose Estimation without Feature Tracking. IEEE Automatic Face and Gesture Recognition, 1998, pp.88-93.
[19] Q. Chen, H. Wu, T. Shioyama, T. Shimada, Kunihiro Chihara. A Robust Algorithm for 3D Head Pose Estimation. IEEE Pattern Recognition, 1998, vol.2,
pp.1356-1359.
[20]W. Niem. Robust and Fast Modelling of 3D Natural Objects from Multiple Views. Proceedings of SPIE, February 1994.
[21]W. Niem, M. Steinmetz. Camera Viewpoint Control for the Automatic Reconstruction of 3D Objects. IEEE Image Processing, 1996, vol.3, pp655-658.
[22]W. Niem, J. Wingbermuhle. Automatic Reconstruction of 3D Object Using a Mobile Monoscopic Camera. IEEE 3-D Digital Imageing and Modeling, 1997, pp.173-180.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top