臺灣博碩士論文加值系統

　　在生物科技的應用中，共軛焦雷射掃描顯微鏡(Confocal Laser Scanning Microscope)所取得之切片影像疊(Slice Image Stack)，能夠在X、Y軸得到極高解析度的影像，但受到光學的限制，在Z軸方向，只能以較X、Y軸為差的解析度進行切片掃描，因此，若能以不同位置下X、Y軸的高解析度影像補足原本Z軸影像的不足，便能取得三軸皆有高解析度的影像模型。
　　利用動機所設計出一嶄新的可動平台機構，由其所取得不同位置下的切片影像疊，設計演算法將這些切片影像疊對齊，以利後續處理。
　　首先對不同位置的切片影像疊進行特徵點擷取，並比對尋找可能的特徵配對，用以計算轉換矩陣(Transformation Matrix)並對不同位置切片影像疊做初步對齊。最後，再將已初步對齊的切片影像疊利用最佳化方法，作最後的對齊修正，取得切片影像疊的精準對位。

In biotechnology applications, confocal laser scanning microscope (CLSM) can obtain slice image stacks with high resolution on the X and Y-axis, but rather poor on the Z-axis because of optical reasons. Thus, if we can obtain image stacks from different positions of the same object, the gap on the Z-axis can be filled in with new X and Y-axis information provided by the different stacks.
The objective is to design an image registration method to collaborate with a newly designed adjustable stage which was invented by Nano-Engineering Institute of Tsing-Hua University.
Our first step is to match feature points from these images, then minimize the error between them. Finally, we use an optimization algorithm to get the best registration position. Thus, we can have the optimal rigid transformation matrix.

Table of Contents
Abstract i
Table of Contents ii
List of Figures iv
Chapter 1. Introduction 1
1.1. Confocal Laser Scanning Microscope 1
1.2. Motivation 3
1.3. Thesis Organization 4
Chapter 2. Super-Resolution 3-D Imaging 5
2.1. Deconvolution 6
2.2. Registration 7
2.3. Interpolation 8
Chapter 3. Related Work 9
3.1. Iterative Closest Point 9
3.1.1. Quaternion Representation of Rotation 10
3.1.2. Solving the Objective Function 10
3.1.3. Iterative Solving Process 12
3.2. Object Recognition 14
3.2.1. Edge Detection 14
3.2.2. Blob Detection 15
Chapter 4. Initial Stack Registration Using SIFT 16
4.1. Feature Extraction 19
4.1.1. Parameter Selection 19
4.1.2. SIFT Frames, Descriptors, and Matching 21
4.2. Handling of Erroneous Match 24
4.3. Generating Transformation Matrix 26
Chapter 5. Particle Swarm Optimization for Optimal Registration 28
5.1. Local Scale Registration 29
5.1.1. Calculating Line Vector 31
5.2. Particle Swarm Optimization 32
5.2.1. Algorithm Description 33
5.2.2. Registration Application 36
Chapter 6. Experimental Results 49
6.1. SIFT Registration Result 49
6.2. PSO Registration Result 53
Chapter 7. Conclusion and Future Work 61
7.1. Conclusion 61
7.2. Future Work 62
Reference 64

Reference
[1] James B. Pawley, “Handbook of biological confocal microscopy 2nd,” Plenum Press.
[2] http://en.wikipedia.org/wiki/Confocal_laser_scanning_microscopy
[3] P.J. Besl and H. D. McKay, “A method for registration of 3-D shapes, “Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 14, pp. 239-256, 1992.
[4] B. K. P. Horn, “Closed-Form Solution of Absolute Orientation Using Unit Quaternions, “J. Opt. Soc. Am. A, vol. 4, pp. 629-642, 1987.
[5] Sobel, I., Feldman,G., “A 3x3 Isotropic Gradient Operator for Image Processing,” presented at a talk at the Stanford Artificial Project in 1968, unpublished but often cited, orig. in Pattern Classification and Scene Analysis, Duda,R. and Hart,P., John Wiley and Sons,'73, pp271-2.
[6] Canny, J., “A Computational Approach to Edge Detection,” IEEE Trans. Pattern Analysis and Machine Intelligence, 8(6):679–698, 1986.
[7] D. G. Lowe, “Object Recognition from Local Scale-Invariant Features,“ in Computer Vision, 1999. The Proceedings of the seventh IEEE International Conference on, 1999, pp. 1150-1157 vol.2.
[8] D. G. Lowe, “Distinctive Image Features from Scale-Invariant Keypoints, “Int. J. Comput. Vision, vol. 60, pp. 91-110, 2004.
[9] W. Kabsch, “A Solution for The Best Rotation to Relate Two Sets of Vectors, “Acta Crystallographica Section A, vol.32, pp. 922-923, 1976.
[10] J. Kennedy and R. Eberhart, “Particle Swarm Optimization, “in Neural Networks, 1995. Proceedings., IEEE International Conference on, 1995, pp. 1942-1948 vol.4.
[11] http://en.wikipedia.org/wiki/Particle_swarm_optimization