零件分類為自動化產線上主要的工作之一，透過機器視覺結合機器學習的技術，可以快速辨識產線上的各種零件，然而，機器學習需要龐大數據進行訓練，同時也會消耗大量的訓練時間，對於少量多樣的產線非常不利。為了克服此問題，本論文以少量的樣品進行零件的辨識，僅透過影像處理技術即可迅速完成零件分類的工作，此外本論文以拼圖做為主要的探討對象，其原因是拼圖必須同時考慮輪廓、顏色、位置、旋轉、辨識時間及組裝等問題，相較一般工業零件僅考慮輪廓外型，拼圖的辨識困難許多。
在自動化拼圖的前置作業中，本論文將完整拼圖進行切割，得到局部拼圖特徵的目標資料，接著使用相機拍攝散落的拼圖，並擷取每一片拼圖的輪廓，以利獲得拼圖片的資料，然後將拼圖片與目標資料進行匹配，匹配過程中將影像進行極座標轉換，減少匹配計算所消耗的時間，並同時找出拼圖片的編號及角度，最後將匹配結果結合機械手臂完成拼圖的組裝。
本論文除了簡述如何透過影像處理產生拼圖片資料以及拼圖的目標資料，也詳述在影像匹配中如何得到對應編號以及角度，最終以不同種類的拼圖驗證此方法的可行性，以及套用至工業上的零件驗證此系統的實用性。

Classification of components is one of the primary tasks in automated production lines. By leveraging machine vision combined with machine learning techniques, various components on the production line can be rapidly identified. However, machine learning requires vast amounts of data for training and consumes significant training time, which poses challenges for small-scale, diverse production lines. To overcome this issue, this study proposes a method for component recognition using a small number of samples, solely relying on image processing techniques to accomplish the task of component classification quickly. Furthermore, this thesis focuses on puzzles as the main subject of investigation. Puzzles present additional challenges compared to typical industrial components since their recognition involves considerations such as contours, colors, positions, rotations, recognition time, and assembly. In contrast, general industrial components often only require contour-based recognition, making puzzle recognition much more difficult.
In the preliminary stages of automated puzzle assembly, this thesis proposes a segmentation approach to divide the puzzle into individual pieces with partial features, obtaining target data for each puzzle piece. Subsequently, scattered puzzle pieces are captured using a camera, and their contours are extracted to acquire puzzle image data. The puzzle images are then matched with the target data, employing polar coordinate transformation during the matching process to reduce computational time. The matching process also identifies the puzzle piece's number and orientation. Finally, the matching results are integrated with a robotic arm to complete the puzzle assembly.
This thesis not only outlines the generation of puzzle image data and target data through image processing but also elaborates on the determination of corresponding numbers and orientations during image matching. The feasibility of this method is validated through different types of puzzles, and its applicability to industrial components is examined for practicality

[1] Roger W. Webster, Paul S. LaFollette and Paul S. LaFollette, “Isthmus Critical Points for Solving Jigsaw Puzzles in Computer Vision,” IEEE Transaction of Systems, Man and Cybernetics, Vol. 21, No. 5, pp. 1271-1278, 1911.
[2] David A. Kosiba, Pierre M. Devaux, Sanjay Balasubramanian, Tarak L. Gandhi and Rangachar Kasturi “An Automatic Jigsaw Puzzle Solver,” IEEE Transaction of Pattern Recognition, Vol. 1, pp. 616-618, 1994.
[3] Feng-Hui Yao and Gui-Feng Shao “A Shape and Image Merging Technique to Solve Jigsaw Puzzles,” Pattern Recognition, pp. 1819-1835, 2003.
[4] Miri Weiss-Cohen and Yoram Halevi “Knowledge Retrieval for Automatic Solving of Jigsaw Puzzles,” IEEE Transaction of Computational Intelligence for Modeling Control and Automation, Vol. 2, pp. 379-383, 2005.
[5] Taeg Sang Cho, Shai Avidan and William T. Freeman “A probabilistic image jigsaw puzzle solver,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2010.
[6] Dolev Pomeranz, Michal Shemesh, and Ohad Ben-Shahar “A Fully Automated Greedy Square Jigsaw Puzzle Solver,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2011.
[7] Xingwei Yang, Nagesh Adluru, and Longin Jan Latecki “Particle Filter with State Permutations for Solving Image Jigsaw Puzzles,” Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2011.
[8] Andrew C. Gallagher “Jigsaw Puzzles with Pieces of Unknown Orientation,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 382-389, 2012.
[9] Dror Sholomon, Omid E. David and Nathan S. Netanyahu “A Generalized Genetic Algorithm-Based Solver for Very Large Jigsaw Puzzles of Complex Types,” Proceedings of the Twenty-Eighth AAAI Conference on Artificial Intelligence, July 27-31, Qu´ebec City, Qu´ebec, Canada., pp. 2839-2845, 2014.
[10] 林奇麟，「結合機器視覺技術應用於 自動化拼圖組裝」，碩士論文，國立臺灣海洋大學機械與機電工程學系，2007。
[11] Quang-Cherng Hsu, Ngoc-Vu Ngo and Rui-Hong Ni “Development of A Faster Classification System for Metal Parts Using Machine Vision Under Different Lighting Environments,” The International Journal of Advanced Manufacturing Technology, Vol. 100, Issue 9-12, pp. 3219-3235, 2019.
[12] Sondos M. Fadl, Noura A. Semary “Robust Copy–Move Forgery Revealing in Digital Images Using Polar Coordinate System,” Neurocomputing, Vol. 265, pp. 57-65, 2017.
[13] Tohari Ahmad, Jiankun Hu, Song Wang “Pair-Polar Coordinate-Based Cancelable Fingerprint Templates,” Pattern Recognition, Vol. 44, pp. 2555-2564, 2011.
[14] Yi Zhang, Kai Lu, Yinghui Gao, Kai Xu “A Novel Quantum Representation for Log-Polar Images,” Quantum Inf Process, Vol.12, pp. 3103-3126, 2013.
[15] 張哲瑋，「利用Demon registration及極座標轉換之自動顱骨缺損修復修補演算法」，碩士論文，東海大學工業工程與經營資訊學系，2017。
[16] 許哲魁，「影像相似度評估函數之分析與比較」，碩士論文，國立交通大學資訊工程系，2002。
[17] Xueli Peng, Ruofei Zhong, Zhen Li and Qingyang Li “Optical Remote Sensing Image Change Detection Based on Attention Mechanism and Image Difference,” IEEE Transactions on Geoscience and Remote Sensing, Vol. 59, No. 9, pp. 7296-7307, 2021.
[18] Kai Briechle and Uwe Hanebeck “Template Matching Using Fast Normalized Cross Correlation,” Proceedings of SPIE, Vol. 4387, pp. 95-102, 2001.
[19] Ni Fei, Fu Zhuang, Liu Renqiang, Cao Qixin and Zhao Yanzheng “An Image Processing Approach for Jigsaw Puzzle Assembly,” Emerald Aseembly Automation, Vol. 27, No. 1, pp. 25-30, 2007.
[20] Jiwu Wang, Masanori Sugisaka “Camera Calibration for A Mobile Robot Prototype,” Artificial Life and Robotics, Vol. 7, pp. 91-94, 2003.
[21] Zhang Xiaowen, Ren Yongfeng, Zhen Guoyong, Shan Yanhu, Chu Chengqun and Liang Fan “Camera Calibration Method for Solid Spheres Based on Triangular Primitives,” Precision Engineering, Vol. 65, pp. 91-103, 2020.
[22] Zhuang Zhang, Rujin Zhao, Enhai Liu, Kun Yan and Yuebo Ma “A Single-Image Linear Calibration Method for Camera,” Measurement, Vol. 130, pp. 298-305, 2018.
[23] Zhen Liu, Qun Wu, Suining Wu and Xiao Pan “Flexible and Accurate Camera Calibration Using Grid Spherical Images,” Optics Express, Vol. 25, No. 13, pp. 15269-15285, 2017.
[24] Grigore C. Burdea and Haim J. Wolfson “Solving Jigsaw Puzzles by A Robot,” IEEE Transactions on Robotics and Automation, Vol. 5, No. 6, pp. 752-764, 1989.
[25] Open Source Computer Vision Library (OpenCV), Retrieved from https://opencv.org/