臺灣博碩士論文加值系統

監視系統近年來非常受到大眾們使用，例如：道路交通流量、意外事故、違規事件、機車改裝噪音，在這大範圍的環境之中，絕大部分都是透過警察肉眼取締，或是監視上百支監視器影像及民眾檢舉，卻沒辦法有效率提升違規取締，未來影像辨識，可透過機器學習反覆訓練和預測，系統使用深度學習模組，不斷強化影像辨識準確率，能正確地辨識定義機車車牌。因此本論文利用人工神經網路特性(Artificial Neural Network, ANN)，和更快速區域卷積神經網路(Faster Region-based Convolutional Neural Networks, Faster-RCNN)，經由許多繁雜的神經元傳遞和接收資訊，可以分析圖片中機車的車牌、文字與數字等類別。蒐集道路架設攝影機影像，相當於會有不同角度機車車牌，使用深度學習影像辨識系統，經由實驗結果顯示準確率達到89%。

Surveillance systems have become an important part of the public sphere, monitoring traffic flow, accidents, legal infractions, noise pollution caused by modified motorcycles, etc. The macroscopic approach to policing primarily relies on law enforcement officials identifying offenses with the naked eye, or the visuals captured by hundreds of surveillance cameras and public vigilance, an inefficient combination that stymies civil infraction citation rate. Future visual recognition will be based on machine learning through repetitive training and prediction. The deep learning module of surveillance systems will continuously improve recognition success rate, such as the precise definition of motorcycles in surveillance footage and whether the rider is in compliance with regulations. Therefore this paper applies the properties of Artificial Neural Network (ANN) and Faster Region-Based Convolutional Neural Networks (Faster-RCNN) to transmit and receive complex neural networks for the analysis of motorcycle license plates, along with its alphabetic and numeral information. A collage of visuals captured from various roadside surveillance cameras is the equivalent of a spectrum of angles on the same license plate of a motorcycle. When the images were processed with this deep learning visual recognition system, the experiment achieved a 89% detection rate.

摘 要 I
Abstract II
目 錄 IV
圖目錄 VI
表目錄 VIII
第一章 序論 1
1.1 研究動機與背景 1
1.2 文獻探討 2
1.3 論文大綱 6
第二章 研究架構 7
2.1卷積神經網路 7
2.2深度神經網路架構 9
2.3更快速的區域卷積神經網路 10
2.4系統訓練架構流程 12
2.5偵測系統流程 13
第三章 背景知識 15
3.1監督學習 15
3.2無監督學習 18
3.3半監督學習 19
3.4增強學習 22
3.5區域卷積網路 24
3.6快速區域卷積神經網路 25
3.7 YOLO 25
3.8 YOLOv2 27
第四章 前置作業 28
4.1資料收集 28
4.2圖片標註 29
4.3製作訓練檔案 32
4.4訓練模組 32
4.5模組測試 34
4.6研究方法 35
4.7車牌偵測 36
4.8文字辨識 37
4.9字母重疊處理 37
4.10缺字計算 38
4.11填補缺字 39
4.12積分影像 40
4.13篩選車牌號碼 42
4.14車牌相似字 42
4.15七碼車牌規格判斷方式 43
4.16六碼車牌規格判斷方式 44
第五章 實驗結果與數據 46
5.1 實驗所需軟硬體規格 46
5.2 機車圖片資料 46
5.3 準確率分析 47
5.4 速度分析 49
5.5 實驗結果 49
第六章 結論與未來研究發展 51
6.1 結論 51
6.2 未來研究發展 51
參考文獻 52

[1] M. Pak, and S. Kim, "A Review of Deep Learning in Image Recognition," 4th International Conference on Computer Applications and Information Processing Technology (CAIPT),pp. 1-3,2017.
[2] L. A. Marina, F. D. Moldoveanu, and S. M. Grigorescu, "Environment perception in racing simulators using Deep Neural Networks," International Conference on Optimization of Electrical and Electronic Equipment (OPTIM) & Intl Aegean Conference on Electrical Machines and Power Electronics (ACEMP),pp. 1102-1107, 2017.
[3] H. Jeong, S. Yun, K. M. Yi, and J. Y. Choi, "Category Attentional Search for Fast Object Detection by Mimicking Human Visual Perception," IEEE Winter Conference on Applications of Computer Vision,pp. 829-836, 2015.
[4] M. Manana, C. Tu, and P. A. Owolawi, "A survey on vehicle detection based on convolution neural networks," 3rd IEEE International Conference on Computer and Communications (ICCC),pp. 1751-1755,2017.
[5] K. Huang, and B. Zhang, "Fine-grained Vehicle Recognition by Deep Convolutional Neural Network," 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI),pp. 465-470, 2016.
[6] J.-W Hsieh, S.-H Yu, and Y.-S. chen, " Morphology-based license plate detection from complex scenes," Object recognition supported by user interaction for service robots,pp. 176-179,2002.
[7] Z. Selmi, M. B. Halima, and A. M. Alimi, "Deep Learning System for Automatic License Plate Detection and Recognition," 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), vol. 01,pp. 1132-1138,2017.
[8] R. Girshick, "Fast R-CNN," OCEANS MTS/IEEE Washington,pp. 1-5, 2015.
[9] K. Wang, Y. Dong, H. Bai,and, Y. Zhao, and K. Hu, "Use Fast R-CNN and Cascade Structure for Face Detection," Visual Communications and Image Processing (VCIP),pp. 1-4, 2016.
[10] J. Mahmoodi, "Digit Recognition of Iranian License Plate Based on SOFM and Naive Bayesian Classifier," International Conference on Radar, Antenna, Microwave, Electronics, and Telecommunications (ICRAMET),pp. 36-41, 2017.
[11] H.-J Jeong, K.-S Park, and Y.-G. Ha, " Image Preprocessing for Efficient Training of YOLO Deep Learning Networks," IEEE International Conference on Big Data and Smart Computing (BigComp),pp. 635-637,2018.
[12] J. Redmon, S. Divvala, R. Girshick and A. Farhadi, " You Only Look Once: Unified, Real-Time Object Detection," IEEE Conference on Computer Vision and Pattern Recognition (CVPR),pp. 779-788,2016.
[13] M. Z. Abedin, A. C. Nath, P. Dhar, K. Deb, and M. S. Hossain, "License plate recognition system based on contour properties and deep learning model," IEEE Region 10 Humanitarian Technology Conference (R10-HTC),pp. 590-593, 2017.
[14] T. Bjorklund, A. Fiandrotti, M Annarumma, G. Grancini, and E. Magli, "Automatic License Plate Recognition with Convolutional Neural Networks Trained on Synthetic Data," IEEE 19th International Workshop on Multimedia Signal Processing (MMSP),pp. 1-6, 2017.
[15] F. D. Kurpiel, R. Minetto, and B. T. Nassu, "Convolutional neural networks for license plate detection in images," Image Processing (ICIP),IEEE International Conference on,pp. 3396-3399, 2017.
[16] S. M. Silva, and C. R. Jung, "Real-Time Brazilian License Plate Detection and Recognition Using Deep Convolutional Neural Networks," 30th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI),pp. 55-62, 2017.
[17] G.-S J Hsu, and C.-W. Chiu, Y. Zhao, and K. Hu, "A comparison study on real-time tracking motorcycle license plates," IEEE 12th Image, Video, and Multidimensional Signal Processing Workshop (IVMSP),pp. 1-5,2016.
[18] M.-S. Roth, and J.-Y. Lee, "Refining faster-RCNN for accurate object detection, " Fifteenth IAPR International Conference on Machine Vision Applications (MVA),pp. 514-517, 2017.
[19] G.-S. Hsu, S.-D. Zeng, C.-W. Chiu, and S-L. Chung, "A comparison study on motorcycle license plate detection," IEEE International Conference on Multimedia & Expo Workshops (ICMEW),pp. 1-6, 2015.
[20] C. Cui, N. Sang, C. Gao, and L. Zou, "Vehicle re-identification by fusing multiple deep neural networks, " Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA),pp. 1-6, 2017
[21] S. G. Kim, H. G. Jeon, and H. I. Koo, "Deep-learning-based license plate detection method using vehicle region extraction", Electronics Letters, vol. 53,no.15,pp. 1034-1036, July 2017.
[22] R. Polisgetty, M. Roopaei, and P. Rad, "A Next-Generation Secure Cloud-Based Deep Learning License Plate Recognition for Smart Cities," 15th IEEE International Conference on Machine Learning and Applications (ICMLA),pp. 286-293, 2016.
[23] F. Zhu, Y. Lu, N. Ying, and G. Giakos, "Fast vehicle detection based on evolving convolutional neural network," IEEE International Conference on Imaging Systems and Techniques (IST),pp. 1-4, 2017.
[24] K. Shi, H. Bao, and N. Ma, "Forward Vehicle Detection Based on Incremental Learning and Fast R-CNN," 13th International Conference on Computational Intelligence and Security (CIS),pp. 73-76, 2017.
[25] J. T. Lee, and Y. Chung, "Deep Learning-Based Vehicle Classification Using an Ensemble of Local Expert and Global Networks," IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW),pp. 920-925, 2017.