隨著測量及空間資訊技術的快速發展，與定位有關之研究與應用越發受到人們的重視。在室外環境中，人們可藉由全球衛星定位系統便利地、快速地取得定位服務。在室內環境時，過去常用基於航位推算與無線訊號的定位方式，但是會面臨累積誤差與訊號干擾等問題，其定位問題仍有很大的改進空間。因此我們思考另一方向，使用影像去實現定位服務，其主要概念先建立室內場域影像之模型與其座標資訊，透過影像特徵值匹配判斷自身的位置。我們提出基於深度類神經網路的影像定位技術，卷積神經網路能夠感知影像的局部區域，發現其中高分辨率的局部特徵，並以此特徵構成人類視覺的基礎，成為提高定位識別率的有效手段。我們基於PoseNet之架構，使用23層卷積神經網路架構，在訓練階段前，設置不同尺寸的輸入影像，訓練端到端的位置識別任務，推算相機初始的三維位置向量，再基於拍攝角度進行位置預測的誤差修正。實驗數據採用地下停車場以及故宮南院場域，地下停車場的場景單調且無紋理，而故宮南院光線昏暗且有玻璃反光的現象，這些嚴峻的環境對於卷積神經網路皆是一大挑戰。我們將分析與探討不同影像尺寸對於影像定位的誤差，其初步實驗成果顯示我們設計的新方法能夠有效提升約二至三成的室內定位精度。另外我們也探討場域大小、不同手機平台、誤差修正對於類神經的定位精度，以求得更加精確的位置定位。初步實驗成果顯示我們設計的角度誤差修正法能夠有效提升約二成。

With the rapid development of surveying and spatial information technologies, more and more attentions have been paid to the research and application of positioning. In outdoor environments, people can easily and quickly obtain positioning services through the Global Positioning System (GPS). In indoor environments, the positioning method based on dead reckoning and wireless signals was commonly used in the past, but it will face the problems of cumulative error and signal interference. There is still much space for improvement in positioning problems. Therefore, we think about another idea using images to achieve positioning services. The main concept is to establish the model of indoor field image and its coordinate information, and to judge its position by image eigenvalue matching. We propose a deep neural network based image registration technology. Convolutional Neural Network (CNN) can perceive local areas of images and find high-resolution local features, which form the basis of human sight and become an effective way to improve the identification rate of positioning. Based on the architecture of PoseNet, we use a 23-layer convolutional neural network to set various image size to input to the CNN architecture before the training stage, train end-to-end location identification tasks, and regress the three-dimensional position vector of the camera. The experimental data are from the field of the Southern Branch of the National Palace Museum. We will analyze and discuss the errors of different image sizes in image registration. The preliminary experimental results show that the new method designed by us can effectively improve the accuracy of indoor positioning by about 20 to 30%. In addition, we also discuss the position accuracy of the field size, different mobile phone platforms and error correction for the system of neural network in order to obtain more accurate position. The preliminary experimental results show that the angle error correction method designed by us can effectively improve about 20%.

[1] Sameer Agarwal, Yasutaka Furukawa, Noah Snavely, Ian Simon, Brian Curless, Steven M. Seitz and Richard Szeliski, “Building Rome in A Day,” Communications of the ACM, Vol. 54, No. 10, pp. 105-112, 2011.
[2] Imran Ashraf, Soojung Hur and Yongwan Park, “Application of Deep Convolutional Neural Networks and Smartphone Sensors for Indoor Localization,” Applied Scienc-es, Vol. 9, No. 11, p. 2337, 2019.
[3] Yoshua Bengio, Aaron Courville and Pascal Vincent, “Representation learning: A Review and New Perspectives,” IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), Vol. 35, No. 8, pp. 1798-1828, 2013.
[4] Klemen Bregar and Mihael Mohorčič, “Improving Indoor Localization Using Convo-lutional Neural Networks on Computationally Restricted Devices,” IEEE Access, Vol. 6, pp. 17429-17441, 2018.
[5] Jakob Engel, Thomas Schöps, and Daniel Cremers, “LSD-SLAM: Large-Scale Direct Monocular SLAM,” European Conference on Computer Vision (ECCV). Springer, Cham, pp. 834-849, 2014.
[6] Yasutaka Furukawa, Brian Curless, Steven M. Seitz and Richard Szeliski, “Towards Internet-Scale Multi-View Stereo,” 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1434-1441, 2010.
[7] Clément Godard, Oisin Mac Aodha and Gabriel J. Brostow, “Unsupervised Monocu-lar Depth Estimation with Left-Right Consistency,” Proceedings of the IEEE Confer-ence on Computer Vision and Pattern Recognition (CVPR), pp. 270-279, 2017.
[8] Uwe Grossmann, Sebastian Gansemer and Oliver Suttorp, “RSSI-Based WLAN Indoor Positioning Used within A Digital Museum Guide,” International Journal of Computing, Vol. 7, No. 2, pp. 66-72, 2014.
[9] Qiang Hao, Rui Cai, Zhiwei Li, Lei Zhang, Yanwei Pang and Feng Wu, “3D Visual Phrases for Landmark Recognition,” 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3594-3601, 2012.
[10] Jiheon Kang, Joonbeom Lee and Doo-Seop Eom, “Smartphone-Based Traveled Dis-tance Estimation Using Individual Walking Patterns for Indoor Localiza-tion,” Sensors, Vol. 18, No. 9, p. 3149, 2018.
[11] Alex Kendall and Roberto Cipolla, “Modelling Uncertainty in Deep Learning for Camera Relocalization,” 2016 IEEE International Conference on Robotics and Au-tomation (ICRA), pp. 4762-4769, 2016.
[12] Alex Kendall and Roberto Cipolla, “Geometric Loss Functions for Camera Pose Re-gression with Deep Learning,” Proceedings of the IEEE Conference on Computer Vi-sion and Pattern Recognition (CVPR), pp. 5974-5983, 2017.
[13] Alex Kendall, Matthew Grimes and Roberto Cipolla, “PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization,” Proceedings of the IEEE International Conference on Computer Vision (ICCV), pp. 2938-2946, 2015.
[14] Kun-Chan Lan and Wen-Yuah Shih, “An Indoor Locationtracking System for Smart Parking,” International Journal of Parallel, Emergent and Distributed Systems, Vol. 29, No. 3, pp. 215-238, 2014.
[15] Fan Li, Chunshui Zhao, Guanzhong Ding, Jian Gong, Chenxing Liu and Feng Zhao, “A Reliable and Accurate Indoor Localization Method Using Phone Inertial Sensors,” Proceedings of the 2012 ACM Conference on Ubiquitous Computing, pp. 421-430, 2012.
[16] Jason-Zhi Liang, Nicholas Corso, Eric Turner and Avideh Zakhor, “Image-Based Positioning of Mobile Devices in Indoor Environments,” Multimodal Location Estimation of Videos and Images, Springer, Cham, pp. 85-99, 2015.
[17] Ayush Mittal, Saideep Tiku and Sudeep Pasricha, “Adapting Convolutional Neural Networks for Indoor Localization with Smart Mobile Devices,” Proceedings of the 2018 on Great Lakes Symposium on VLSI, ACM, pp. 117-122, 2018.
[18] Andrew Ng, Jiquan Ngiam, Chuan-Yu Foo, Yifan Mai, Caroline Suen, Adam Coates, Andrew Maas, Awni Hannun, Brody Huval, Tao Wang and Sameep Tandon, “Convolutional Neural Networks,” http://ufldl.stanford.edu/tutorial/supervised/ConvolutionalNeuralNetwork/, 2013.
[19] Arne Niitsoo, Thorsten Edelhäuβer and Christopher Mutschler, “Convolutional Neu-ral Networks for Position Estimation in TDoA-Based Locating Systems,” 2018 Inter-national Conference on Indoor Positioning and Indoor Navigation (IPIN), IEEE, pp. 1-8, 2018.
[20] Maxime Oquab, Leon Bottou, Ivan Laptev and Josef Sivic, “Learning and Transferring Mid-Level Image Representations Using Convolutional Neural Networks,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1717-1724, 2014.
[21] Ali-Sharif Razavian, Hossein Azizpour, Josephine Sullivan and Stefan Carlsson, “CNN Features Off-the-Shelf: An Astounding Baseline for Recognition,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 806-813, 2014.
[22] Jamie Shotton, Ben Glocker, Christopher Zach, Shahram Izadi, Antonio Criminisi and Andrew Fitzgibbon, “Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2930-2937, 2013.
[23] Noah Snavely, Steven M. Seitz and Richard Szeliski, “Photo Tourism: Exploring Photo Collections in 3D,” ACM Transactions on Graphics (TOG), Vol. 25, No. 3, pp. 835-846, 2006.
[24] Fazli Subhan, Halabi Hasbullah, Azat Rozyyev and Sheikh-Tahir Bakhsh, “Indoor Positioning in Bluetooth Networks Using Fingerprinting and Lateration Approach,” 2011 International Conference on Information Science and Applications (ICISA), IEEE, pp. 1-9, 2011.
[25] Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke and Andrew Rabinovich, “Going Deeper with Convolutions,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1-9, 2015.
[26] Florian Walch, Caner Hazirbas, Laura Leal-Taixé, Torsten Sattler, Sebastian Hilsenbeck and Daniel Cremers, “Image-Based Localization Using LSTMs for Struc-tured Feature Correlation,” Proceedings of the IEEE International Conference on Computer Vision (ICCV), pp. 627-637, 2017.
[27] Xuyu Wang, Xiangyu Wang and Shiwen Mao, “Deep Convolutional Neural Net-works for Indoor Localization with CSI Images,” IEEE Transactions on Network Science and Engineering, 2018.
[28] Changchang Wu, “Towards Linear-Time Incremental Structure from Motion,” 2013 International Conference on 3D Vision-3DV 2013, IEEE, pp. 127-134, 2013.
[29] Bolei Zhou, Agata Lapedriza, Jianxiong Xiao, Antonio Torralba and Aude Oliva, “Learning Deep Features for Scene Recognition Using Places Database,” Proceedings of the 27th International Conference on Neural Information Processing Systems, Vol. 1, pp. 487-495, 2014.
[30] Tinghui Zhou, Matthew Brown, Noah Snavely and David G. Lowe, “Unsupervised Learning of Depth and Ego-Motion from Video,” Proceedings of the IEEE Confer-ence on Computer Vision and Pattern Recognition (CVPR), pp. 1851-1858, 2017.
[31] 江凱偉, 曾義星 and 呂學展, “107年度移動載台測量製圖技術發展工作案期末報告,” 內政部地政司, 2018.