跳到主要內容

臺灣博碩士論文加值系統

(44.200.82.149) 您好!臺灣時間:2023/06/11 01:37
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:郭文筱
研究生(外文):Wen-Hsiao Kuo
論文名稱:針對未校正影像序列作影像對應之研究
論文名稱(外文):A Study on Image Matching for Uncalibrated Image Sequences
指導教授:唐政元
指導教授(外文):Cheng-Yuan Tang
學位類別:碩士
校院名稱:華梵大學
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:2005
畢業學年度:93
語文別:中文
中文關鍵詞:立體對應未校正影像Kernel-based Color HistogramHausdorff Distance
外文關鍵詞:Stereo correspondenceUncalibrated imagesKernel-based Color HistogramHausdorff Distance
相關次數:
  • 被引用被引用:2
  • 點閱點閱:302
  • 評分評分:
  • 下載下載:35
  • 收藏至我的研究室書目清單書目收藏:2
本篇論文目的是在未校正的影像序列上,求出影像特徵點在影像間的對應關係。在本論文中,我們把影像對應的問題分為兩個部分:一個是利用影像平面上的資訊來求得對應點;另一個是透過三維資訊將錯誤的對應點刪除,且設法找出更多的對應點。
首先,我們提出利用核心函式(Kernel function)結合Hausdorff Distance及Color Histogram的方法將影像的對應點求出。並且針對我們所改良的Kernel-based Hausdorff Distance及Kernel-based Color Histogram,與前人所提出的對應方法,例如:Normalized Cross-Correlation (NCC)、Sum of Squared Differences (SSD),以及Rank、Census transform作比較分析。經由實驗結果比較分析後,我們發現:以SSD所求出的對應點,其正確率較其它的方法高。
在以二維對應點估測出基礎矩陣(Fundamental matrix)之後,我們可以求出兩張影像的投影矩陣及對應點的三維座標,並利用這些三維資訊建立更正確的對應點。在本論文中,利用三維資訊的方法分為兩個部分,一個是利用三維資訊將錯誤的對應點刪除;另一個部分是利用三維資訊及其共平面的關係找出更多的對應點。根據我們的實驗,沒有使用三維資訊所找出的對應點正確率為88%;使用三維資訊後,結合SSD和距離(Distance)的方法能將錯誤的對應點刪除,且較不會將正確的對應點刪掉,正確率可達93%。因此,由實驗結果可以發現使用SSD和距離的方法能取得最好的對應點。
The purpose of this paper is to find the corresponding relation from uncalibrated image sequences. We divide the content of our research into two parts. First, we use the 2D information on the image to find the 2D corresponding point pairs. Second, we use the 3D information to delete the wrong corresponding point pairs, and then use the 3D information to find more 2D corresponding point pairs.
In this paper, we propose a method combining the kernel function with Hausdorff distance and color histogram to find the 2D corresponding point pairs. Some experiments has been done for comparing our proposed methods with several correspondence algorithms found in the literature: normalized cross-correlation (NCC), sum of squared differences (SSD), rank, census transform. From the experimental results, using SSD to find the 2D corresponding point pairs can achieve the higher correct rate.
After 2D corresponding point pairs have been found, the fundamental matrix can be estimated. Once the fundamental matrix is available, 3D information is useful for establishing more correct 2D correspondence. In this paper, using 3D information to establish stereo correspondence from uncalibrated images was proposed. Our method using 3D information contains two parts. One is using the 3D information to delete the error correspondence; the other part is using the 3D information to find more 2D corresponding points. Some experimental results to demonstrate the performance of our proposed methods are shown in this paper. According to our experiments, the method without 3D information can achieve the correct rate to be 88%, and the correct rate can improve to be more than 90% after using 3D information. The method combining both SSD and Distance can delete more wrong corresponding pairs and delete less correct corresponding pairs. Therefore, this method can achieve best results. Some experimental results to demonstrate the performance of our proposed methods are shown in this paper.
誌 謝 I
摘 要 II
Abstract III
目 錄 V
表 錄 VII
圖 錄 VIII
符號說明 X
一、 緒論 - 1 -
1.1 研究動機 - 1 -
1.2 相關研究 - 2 -
1.2.1 影像辨識 - 2 -
1.3 論文架構 - 4 -
二、 對應的方法 - 5 -
2.1 區塊對應(Block Methods) - 6 -
2.2 Kernel-based Hausdorff Distance - 7 -
2.2.1 Hausdorff Distance - 7 -
2.2.2 Kernel-Based Hausdorff Distance - 9 -
2.3 Kernel-based Color Histogram - 10 -
三、 影像對應之架構 - 13 -
3.1 特徵擷取 - 13 -
3.2 特徵對應 - 15 -
3.2.1 顏色分佈比對 (Matching Color Distribution in Possible Candidates) - 16 -
3.2.2 Neighborhood Constraint - 19 -
3.3 實驗結果 - 20 -
四、 利用三維座標找立體對應 - 23 -
4.1 刪除錯誤的對應點 - 24 -
4.2 找出更多的二維對應點 - 25 -
五、 實驗結果 - 27 -
5.1 比較與分析 - 27 -
5.2 刪除錯誤的對應點 - 29 -
5.3 找到更多的對應點 - 35 -
六、 結論及未來工作 - 37 -
6.1 結論及貢獻 - 37 -
6.2 未來工作 - 38 -
參考文獻 - 41 -
[1]A. Barla, F. Odone and A. Verri. “Hausdorff Kernel for 3D Object Acquisition and Detection,” In European Conference on Computer Vision, 2002, pp. 20-33.
[2]M. Z. Brown, D. Burschka, “Advances in Computational Stereo,” IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 25, No. 8, August 2003, pp. 993-1008.
[3]M. P. Dubuisson, A. K. Jain. “A Modified Hausdorff distance for object matching, ” Proc. of IAPR Int. Conf. on Pattern Recognition (ICPR'94, Jerusalem, IS) , 1994, pp. 566-568.
[4]D. A. Forsyth and J. Ponce. Computer Vision: A Modern Approach, Prentice Hall, 2003.
[5]D. B. Gennery, “Visual Tracking of Known Three Dimensional Objects,” International Journal of Computer Vision, Vol. 7, No. 3, 1992, pp. 243-270.
[6]B. Georgescu, P. Meer, “Point Matching under Large Image Deformations and Illumination Changes,” IEEE Trans. PAMI, Vol. 26, No. 6, June 2004, pp. 674-688.
[7]R. M. Haralick and L. G. Shapiro. Computer and Robot Vision. Vol. I, Addison-Wesley, 1992.
[8]R. M. Haralick and L. G. Shapiro, Computer and Robot Vision. Vol.Ⅱ. Addison-Wesley Publishing Company, Inc., 1993.
[9]C.G. Harris and M.J. Stephens. “A Combined Corner and Edge Detector,” Proceedings Fourth Alvey Vision Conference, Manchester, 1988, pp. 147-151.
[10]R. Hartley, Multiple View Geometry in Computer Vision. Cambridge University Press 2000.
[11]H. C. Huang and Y. P. Hung, “Adaptive Early-Jump-Out Technique for Fast Motion Estimation in Video Coding,” Graphical Model and Image Processing, Vol. 59, No. 6, 1997, pp. 388-394.
[12]D. Huttenlocher, G. Klanderman and W. Rucklidge, “Comparing Images Using the Hausdorff Distance, ” IEEE Trans. PAMI, Vol. 15, No. 9, 1993, pp. 850-863.
[13]F. Odone, A. Barla and A. Verri, “Building Kernels From Binary Strings for Image Matching,” IEEE Trans. Image Processing, Vol. 14, No. 2, February 2005, pp. 169-180.
[14]M. Okutomi and T. Kanade, “A Multiple-Baseline Stereo,” IEEE Trans. PAMI, Vol. 15, No. 4, 1993, pp. 353-363.
[15]B. Scholkopf and A. J. Smola. Learning with Kernels, 2002.
[16]M. J. Swain and D. H. Ballard, "Color Indexing," International Journal of Computer Vision, Vol. 7, No. 1, 1991, pp. 11-32.
[17]C. Y. Tang, H. L. Chou, Y. H. Yeh, M. W. Lin, W. H. Kuo and Y. N. Liao, “Kernel-based Stereo Correspondence for Color Images,” Proc. of CVGIP, 2004.
[18]C. Tomasi and T. Kanade, “Shape and Motion from Image Streams Under Orthography: A Factorization Approach,” International Journal of Computer Vision, Vol. 9, No. 2, 1992, pp. 137-154.
[19]R. Zabih and J. Woodfill, “Non-parametric Local Transforms for Computing Visual Correspondence,” Proc. Third European Conf. Computer Visual, 1994, pp. 150-158.
[20]http://www-cgrl.cs.mcgill.ca/~godfried/teaching/cg-projects/98/normand/main.html
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top