臺灣博碩士論文加值系統

隨著三維模型資料量的增加，使用者在龐大的三維模型資料庫中找出特定的模型變得更加不易，因此與三維模型搜尋的相關研究也日益重要。三維型狀分佈描述(3D Shape Spectrum Descriptor, 3DSSD)，或稱為Shape 3D Descriptor是MPEG-7制定的三維模型搜尋標準描述之一。它呈現了出一個模型中各種三維形狀(如球狀、脊狀或馬鞍狀等)的機率分佈，利用「類似的物體應有相近的三維形狀比例」之概念，來進行三維模型的搜尋。
本篇論文提出一演算法，自動計算出符合MPEG-7標準的描述值，並實際設計出一個三維模型搜尋系統。實作分為兩部分，包含粹取資料庫中模型的三維型狀分佈描述，以及比對粹取出的描述。
我們並對此方法的效能進行測試，以網路上(3D Cafe)下載的1833個三維模型為實驗對象，顯示3DSSD與其他方法的效能之比較，並對此描述作優劣的分析。

With the rapid growth of the amount of 3D models, retrieval of specific content from massive 3D model database has become an important issue in resent years. 3D Shape Spectrum Descriptor (3DSSD), or named Shape 3D Descriptor is a standard set up by MPEG-7 for 3D model retrieval. It is a descriptor that represents the distribution of 3D shapes, such as spherical, ridged or saddle-like ones, on models. It compares models using the concept that “similar objects have similar proportion of the 3D shapes locally”.
In this thesis, we propose an algorithm which is primarily based on MPEG-7 and computes the standard compatible descriptors of 3D models automatically. In addition, we design a system for 3D model retrieval to prove the feasibility of our algorithm. Our system is composed of two parts. The first is to extract the 3DSSD from models in the database, and the second is to compare the descriptors.
We also evaluate the performance of 3DSSD using 1833 models downloaded from the web site “3DCafe”, and compare our results with other descriptions. Finally, we analyze the advantages and the disadvantage of Shape 3D Descriptor.

Contents
中文摘要 2
ABSTRACT 3
致謝 4
CHAPTER 1 INTRODUCTION 9
1.1 3D MODEL DESCRIPTION IN MPEG-7 STANDARD 9
1.2 MOTIVATION 10
1.3 ORGANIZATION 10
CHAPTER 2 RELATED WORK 12
2.1 TEXT QUERY 12
2.2 VISUAL-BASED SIMILARITY 13
2.3 SKELETON MATCHING 15
2.4 3D HARMONICS 15
2.5 CURVATURE-BASED SIMILARITY 16
2.6 LEARNING 16
CHAPTER 3 BACKGROUND 17
3.1 THE PRINCIPAL CURVATURES 17
3.2 SHAPE INDEX AND CURVEDNESS 19
CHAPTER 4 EXTRACTION AND COMPARISON OF 3D SHAPE SPECTRUM DESCRIPTOR 22
4.1 THE EXTRACTION OF 3DSSD 23
4.1.1 The Shape 3D Framework in MPEG-7 23
4.1.1.1 Computation of the mean normal vector 24
4.1.1.2 Local surface approximation with a explicit function 25
4.1.1.3 Principal curvature estimation 26
4.1.2 3D-SS computation 27
4.1.3 Remeshing and Some Modification 28
4.1.4 Previews of the Surface Function Approximation Result 31
4.2 THE COMPARISON AMONG 3DSSDS 33
CHAPTER 5 EXPERIMENTAL RESULTS AND DISCUSSION 36
5.1. RESULTS 36
5.2. PERFORMANCE EVALUATION 42
5.3. DISCUSSION 44
5.3.1 Advantages 44
5.3.2 Disadvantages 47
CHAPTER 6 CONCLUSION AND FUTURE WORK 52
6.1. CONCLUSION 52
6.2. FUTURE WORK 52
REFERENCE 53
RESUME 55

Reference
[1] Ding-Yun Chen, Xiao-Pei Tian, Yu-Te Shen and Ming Ouhyoung, “On Visual Similarity Based 3D Model Retrieval,” to appear in Computer Graphics Forum (EUROGRAPHICS’03), Vol. 22, No. 3, Sep. 2003.
[2] Masaki Hilaga, Yoshihisa Shinagawa, Taku Khomra, and Tosiyasu L. Kunii, “Topology Matching for Full Automatic Similarity Estimation of 3D Shapes,” in Proceedings of SIGGRAPH 2001, pp. 203—212.
[3] T. Funkhouser, P. Vin, M. Kazhdan, J. Chen, A. Halderman, D. Dobkin, and D. Jacobs, “A Search Engine for 3D Models,” ACM Transaction on Graphics, Vol. 22, No. 1, pp. 83—105, Jan. 2003.
[4] T. Zaharia and F. Preteux, “Three-dimensional shape-based retrieval within the MPEG-7 framework,” in Proceedings of SPIE Conference 4304 on Nonlinear Image Processing and Pattern Analysis XII, pp. 133—145, Jan. 2001.
[5] C. Zhang and T. Chen, “An Active Learning Framework for Content Based Information Retrieval,” IEEE Transaction on Multimedia, Special Issue on Multimedia Database, Jun. 2002.
[6] J. J. Koenderink and A. J. van Doorn, “Surface Shape and Curvature Scales,” in Proceedings of IVC, Vol. 10, pp. 557—565, 1992.
[7] A. E. Bimbo, Visual Information Retrieval, Morgan Kaufmann Publishers, Inc., 1999.
[8] 3D Café: a web site of 3d models. http://www.3dcafe.com.
[9] S. Jeannin, L. Cieplinski, J. R. Ohm, and M. Kim, MPEG-7 Visual part of Xperimentation Model Version 7.0, ISO/IEC JTC1/SC29/WG11/N3521, Beijing, China, Jul. 2000.
[10] MPEG-7 Visual part of XM, Version 8.0.
[11] SpharmonicKit 2.5: Fast spherical transforms. http://www.cs.dartmouth.edu/~geelong/sphere/
[12] I. T. Jolliffe, Principal Component Analysis, 2nd edition, Springer, 2002.
[13] Google: a powerful search engine. http://www.google.com
[14] A. McCallum, 1996. Bow: A toolkit for statistical language modeling, text retrieval, classification and clustering. http://www.cs.cmu.edu/˜mccallum/bow
[15] G. A. Miller, “WordNet: A lexical database for English”, Communications of the ACM, Vol. 38, No. 11, pp. 39—41. 1995.
[16] Mathematics, Wolfram Research: http://mathworld.wolfram.com/NormalCurvature.html