隨著工業4.0的發展更加成熟，以及新型冠狀病毒的影響，越來越多的企業願意引進無人化設備，然而，無人化設備數量的增加，也引發了設備管理的問題，因此，本研究主旨在開發一套室內影像定位與追蹤系統，主要分為五個部分: 模組化設備、相機標定、通訊、物件追蹤和3D座標重建。在單一模組化的設備上，具有五個相機模組，分別拍攝中央、前方、後方、右方和左方，彼此間具有150°夾角，當模組架設於高度2800毫米時，透過影像拼接後，可以涵蓋64平方公尺(8公尺*8公尺)，所產生的全景圖解析度為1980像素*1420像素，若使用單一控制器要同時獲取五台相機的資料、辨識和追蹤，對於硬體的運算負擔過重，因此，本研究採用分散式架構，一個相機模組由一個嵌入式設備負責，使用樹莓派4B 8GB作為控制器，除了可以緩解硬體負擔過重的問題外，也可以增加系統的穩定度，避免因為模組內的單一設備故障，而導致整個模組無法正常運作。然而，隨著設備的增加，彼此間的通訊也變得重要，透過將本地端電腦當作伺服器端，輪流詢問每台控制器，是否有辨識到物件，以及物件所在位置，在本系統中採用TCP/IP的通訊架構，並架設於內網中，以避免資料外洩的問題。在單一模組內，可以進行物件的2D追蹤，若同時使用兩組，則可以透過雙目視覺，計算出物件移動的3D軌跡。本研究測試兩種不同的基線長度，第一種為210毫米，且將模組架設於高度2800毫米時，量測出的誤差結果為23.314毫米，而三維軌跡的高度誤差在30毫米以內，另外，第二種為840毫米，且架設於高度2000毫米時，量測誤差皆在3毫米之內，而三維軌跡的高度誤差在10毫米以內。

With the more mature development of Industry 4.0 and the impact of the new coronavirus, more and more companies are willing to introduce unmanned equipment. However, the increase in the number of unmanned equipment has also caused problems in equipment management. Therefore, the main purpose of this research is in the development of an indoor image positioning and tracking system, it is divided into five parts: modular equipment, camera calibration, communication, object tracking and 3D coordinate reconstruction. On a single modular device, there are five camera modules which shoot the center, front, rear, right and left, respectively, with a 150° angle between each other. When the module is erected at a height of 2800 mm, after the image stitching, it can cover 64 m^2 (8m*8 m) and the resolution of the generated panorama is 1980 pixels * 1420 pixels. If a single controller is used to obtain data, identification and tracking of five cameras at the same time, the computational burden on the hardware is too heavy. Therefore, this study adopts a distributed architecture that one camera module is controlled by an embedded device and the use of raspberry Pi 4B 8GB as the controller that not only alleviate the problem of overloading the hardware, but also increase the stability of the system and avoid the failure of a single device in the module which will cause the entire module to fail to operate normally. However, with the increase of devices, the communication with each other also becomes important. By using the local computer as a sever that polling each controller to see whether an object is recognized and the location of the object and set up the user interface that is convenient for users to obtain information and issue commands. In this system that adopts the communication structure of TCP/IP and is set up in the intranet to avoid the problem of data leakage. In a single module, 2D tracking of objects can be performed. If two groups are used at the same time, the 3D trajectory of object movement can be calculated through binocular vision. In this study, two different baseline lengths were tested. The first was 210 mm. When the module was erected at a height of 2800 mm, the measured error was 23.314 mm, while the height error of the 3D trajectory was within 30 mm. In addition, the second type is 840 mm, and when it is erected at a height of 2000 mm, the measurement error is all within 3 mm, and the height error of the three-dimensional trajectory is within 10 mm.

參考文獻
[1] Z. Zhang, "A flexible new technique for camera calibration," IEEE Transactions on pattern analysis and machine intelligence, vol. 22, no. 11, pp. 1330-1334, 2000.
[2] A. Fetić, D. Jurić, and D. Osmanković, "The procedure of a camera calibration using Camera Calibration Toolbox for MATLAB," in 2012 35th International Convention MIPRO, 2012: IEEE, pp. 1752-1757.
[3] J. Kannala and S. S. Brandt, "A generic camera model and calibration method for conventional, wide-angle, and fish-eye lenses," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 8, pp. 1335-1340, 2006, doi: 10.1109/TPAMI.2006.153.
[4] L. Yang, B. Wang, R. Zhang, H. Zhou, and R. Wang, "Analysis on Location Accuracy for the Binocular Stereo Vision System," IEEE Photonics Journal, vol. 10, pp. 1-16, 2018.
[5] A. Bushnevskiy, L. Sorgi, and B. Rosenhahn, "Multimode camera calibration," in 2016 IEEE International Conference on Image Processing (ICIP), 2016, pp. 1165-1169.
[6] Y. Dong, X. Ye, and X. He, "A novel camera calibration method combined with calibration toolbox and genetic algorithm," in 2016 IEEE 11th Conference on Industrial Electronics and Applications (ICIEA), 2016, pp. 1416-1420.
[7] Z. Lin, X. Tang, and H. Wei, "Non-Ghosting Stitching for Street-View Image," in 2020 Cross Strait Radio Science & Wireless Technology Conference (CSRSWTC), 2020, pp. 1-3.
[8] Y. Yuan, F. Fang, and G. Zhang, "Superpixel-Based Seamless Image Stitching for UAV Images," IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 2, pp. 1565-1576, 2021.
[9] C. -C. Lin, S. U. Pankanti, K. N. Ramamurthy, and A. Y. Aravkin, "Adaptive as-natural-as-possible image stitching," in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 1155-1163.
[10] L. Nie, C. Lin, K. Liao, and Y. Zhao, "Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images," IEEE Transactions on Image Processing, vol. 30, pp. 6184-6197, 2021.
[11] A. Garg and L.-R. Dung, "Stitching Strip Determination for Optimal Seamline Search," in International Conference on Imaging, Signal Processing and Communications (ICISPC), 2020, pp. 29-33.
[12] C. Xiu and Y. Ma, "Image Stitching Based on Improved Gradual Fusion Algorithm," in 2019 Chinese Control And Decision Conference (CCDC), 2019, pp. 2933-2937.
[13] H. Bay, A. Ess, T. Tuytelaars, and L. Gool, "SURF: Speeded Up Robust Features," Computer Vision and Image Understanding, vol. 110, no. 3, pp. 346-359, 2008.
[14] E. Rublee, V. Rabaud, K. Konolige, and G. Bradski, "ORB: An efficient alternative to SIFT or SURF," in International Conference on Computer Vision, 2011, pp. 2564-2571.
[15] D. G. Lowe, "Distinctive image features from scale-invariant keypoints," International journal of computer vision, vol. 60, no. 2, pp. 91-110, 2004.
[16] D. Mohanapriya and M. K, "Multi object tracking using gradient-based learning model in video-surveillance," China Communications, vol. 18, no. 10, pp. 169-180, 2021.
[17] Y. Qian, H. Shi, M. Tian, R. Yang, and Y. Duan, "Multiple Object Tracking for Similar, Monotonic Targets," in 2020 10th Institute of Electrical and Electronics Engineers International Conference on Cyber Technology in Automation, Control, and Intelligent Systems (CYBER), 2020, pp. 360-363.
[18] Y. Zhang, Z. Chen, and B. Wei, "A Sport Athlete Object Tracking Based on Deep Sort and Yolo V4 in Case of Camera Movement," in 2020 IEEE 6th International Conference on Computer and Communications (ICCC), 2020, pp. 1312-1316.
[19] H. Li, H. Lin, R. Zhang, L. Lei, D. Wang, and J. Li, "Object Tracking in Video Sequence based on Kalman filter," in 2020 International Conference on Computer Engineering and Intelligent Control (ICCEIC), 2020, pp. 106-110.
[20] P. Zhang, Y. Zhang, and X. Hu, "Fast Multiple Object Tracking Using Relevant Motion Vector," in IEEE 13th International Conference on Computer Research and Development (ICCRD), 2021, pp. 1-5.
[21] Y. Dai, C. Wen, H. Wu, Y. Guo, L. Chen, and C. Wang, "Indoor 3D Human Trajectory Reconstruction Using Surveillance Camera Videos and Point Clouds," IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 4, pp. 2482-2495, 2022.
[22] Q. Yu, B. Wang, and Y. Su, "Object Detection-Tracking Algorithm for Unmanned Surface Vehicles Based on a Radar-Photoelectric System," IEEE Access, vol. 9, pp. 57529-57541, 2021.
[23] G. Swalaganata, Muniri, and Y. Affriyenni, "Moving object tracking using hybrid method," in 2018 International Conference on Information and Communications Technology (ICOIACT), 2018, pp. 607-611.
[24] A. Islam, M. Asikuzzaman, M. O. Khyam, M. Noor-A-Rahim, and M. R. Pickering, "Stereo Vision-Based 3D Positioning and Tracking," IEEE Access, vol. 8, pp. 138771-138787, 2020.
[25] Y. Wei and Y. Xi, "Optimization of 3-D Pose Measurement Method Based on Binocular Vision," IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1-12, 2022.
[26] Taryudi and M.-S. Wang, "3D object pose estimation using stereo vision for object manipulation system," in 2017 International Conference on Applied System Innovation (ICASI), 2017: IEEE, pp. 1532-1535.
[27] B. Li, Y. Zhang, B. Zhao, and H. Shao, "3D-ReConstnet: A Single-View 3D-Object Point Cloud Reconstruction Network," IEEE Access, vol. 8, pp. 83782-83790, 2020.
[28] C. Yang, F. Zhou, and X. Bai, "3D Reconstruction through Measure Based Image Selection," in 2013 Ninth International Conference on Computational Intelligence and Security, 2013, pp. 377-381.
[29] X. Qu, J. Zhao, Y. Sun, and L. Wang, "3D Reconstruction Method Based on Aerial Sequence of UAV," in 2020 International Conference on Virtual Reality and Visualization (ICVRV), 2020, pp. 33-37.
[30] S. Bullinger, C. Bodensteiner, and M. Arens, "Monocular 3D Vehicle Trajectory Reconstruction Using Terrain Shape Constraints," in 2018 21st International Conference on Intelligent Transportation Systems (ITSC), 2018, pp. 1122-1128.