進階搜尋


下載電子全文  
系統識別號 U0026-1304202013300300
論文名稱(中文) 從移動測繪資料自動化萃取交通號誌標誌以建置高精地圖
論文名稱(英文) Automatic Extraction of Traffic Signs and Lights From Mobile Mapping Datasets For HD Maps Generation
校院名稱 成功大學
系所名稱(中) 測量及空間資訊學系
系所名稱(英) Department of Geomatics
學年度 108
學期 2
出版年 109
研究生(中文) 王翊銘
研究生(英文) Yi-Ming Wang
學號 P66071096
學位類別 碩士
語文別 英文
論文頁數 72頁
口試委員 指導教授-曾義星
口試委員-史天元
口試委員-徐百輝
口試委員-張智安
中文關鍵字 高精地圖  點雲  光達  電腦視覺  人工智慧  Mask-RCNN 
英文關鍵字 HD maps  point cloud  LiDAR  computer vision  artificial intelligence  Mask-RCNN 
學科別分類
中文摘要 近年來,自駕車產業之興起,隨之帶動相關產業發展,其中高精地圖為自駕車系統中相當基礎且重要之元素,然而目前大部分之工作流程皆以人工方式完成,此方式將耗費大量人力、時間,而過長的處理程序也將提高錯誤之機率,進而影響自駕車產業之發產,因此本研究提出一套演算法可自動化萃取所需之資訊並建置高精地圖。

本研究中選定交通標誌及交通號誌做為測試標的,演算法之整體步驟可分為四部分。首先為點雲資料之預處理,如透過高程濾波、噪音濾波或強度值濾波,來濾除非標的物之點雲資料,而預處理可提升整體處理之效率及最終成果之精度。第二部分為透過AI模型來辨識影像之屬性,高精地圖中之資料可分為兩大類,其一為幾何資料,其二為屬性資料。近年來AI模型對影像分類或影像辨識之精度有所提升,影像辨識之精度遠高於透過點雲來確認屬性,故採用影像來萃取屬性資料。第三部分為點雲及影像資料之連結,將點雲提供之精準幾何資訊以及影像之屬性資料加以結合。然而目前影像辨識仍未能保證不具有任何錯誤,故在此透過投票機制來確保被賦予在點雲上的屬性資料之正確性。最終之步驟為對該結果進行分類、計算,以求得高精地圖中所需之相關資訊。

透過本研究可瞭解高精地圖之相關概念,如其標準、產製過程及其對自駕車之重要性,以確保演算法產出之成果能符合台灣目前訂定之標準。透過此演算法產出之成果和沙崙場域之高精地圖進行比較後,其精度皆符合目前台灣高精地圖之標準,即三維之精度須小於30公分,而萃取出之數量及屬性正確率高達百分之99。本文之實驗成果證實自動化產製高精地圖之可行性。
英文摘要 In recent years, the rise of the autonomous vehicles (AVs) industry leads many scholars into the field of high definition (HD) maps, and HD maps is a basic but important component for autonomous driving. However, the procedure of generating the HD maps is manual now, and that costs a lot of time and labor, thus that hinder the development of AVs industry. Therefore, this research proposed an algorithm for generating HD maps automatically.

Base on the issue above, in this research two classes of targets, traffic sign and traffic light, were chosen for testing the algorithm. The procedure of the proposed algorithm in this research was divided into four parts. The first was the pre-processing for point cloud data, such as the noise, height, or intensity filter. This step improved the efficiency and accuracy of overall research. The second was the image classification by the AI model, the information of HD maps can be divided into geometry and attribute information. The image classification provided the precise attribute information compared with the point cloud data. The third was the linking of point cloud and image data, the correct attribute and geometry information were combined according to the voting mechanism. The final was the calculation for the needed information in HD maps.

According to the proposed method in this study, the accuracy of the result was higher than the regulation in the Taiwan HD maps standard which is 30 centimeters for three- dimensional accuracy and the accuracy for the attribute and count can reach 99.9 percent. This research stated that the method of generating the HD maps automatically is feasible and accurate, which can assist the development of AVs industry.
論文目次 摘要 I
ABSTRACT II
誌謝 III
LIST OF TABLES VI
LIST OF FIGURES VII
Chapter 1 Introduction 1
1.1 Motivation 1
1.2 Objectives 2
1.3 Research Approach 3
1.4 Thesis Structure 4
Chapter 2 High Definition (HD) Maps 5
2.1 Related concepts in HD Maps 5
2.2 Production of HD Maps 7
2.3 Standard of HD Maps in Taiwan 9
2.3.1 Traffic sign 9
2.3.2 Traffic light 12
Chapter 3 Methodology 18
3.1 Overall Procedure 18
3.2 Point Cloud Pre-processing 20
3.3 AI Image Detection of Traffic Signs and Lights 27
3.4 Linking Point Cloud and Image Data 33
3.4.1 Voting mechanism 35
3.4.2 Projecting point to image 37
3.5 Extraction of the HD maps information 43
Chapter 4 Experiments 49
4.1 Testing Field 49
4.2 Testing data 50
4.3 Referenced HD maps 56
4.4 Test Results 59
4.5 Accuracy Assessment 62
Chapter 5 Conclusion 68
REFERENCE 70
參考文獻 Bram, Geenen, Ali, Nasseri, & Adriaan, Schiphorst. (2020). 2020 Autonomous Vehicle Technology Report. WEVOLVER: https://www.wevolver.com/article/2020.autonomous.vehicle.technology.report
Caltagirone, L., Scheidegger, S., Svensson, L., & Wahde, M. (2017, June). Fast LIDAR-based road detection using fully convolutional neural networks. In 2017 ieee intelligent vehicles symposium (iv) (pp. 1019-1024). IEEE.
Che, E., Jung, J., & Olsen, M. J. (2019). Object recognition, segmentation, and classification of mobile laser scanning point clouds: A state of the art review. Sensors, 19(4), 810.
Ester, M., Kriegel, H. P., Sander, J., & Xu, X. (1996, August). A density-based algorithm for discovering clusters in large spatial databases with noise. In Kdd (Vol. 96, No. 34, pp. 226-231).
Fei-Fei Li, Justin Johnson, & Serena Yeung (2017,May). Detection and Segmentation.
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587).
Girshick, R. (2015). Fast r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 1440-1448).
HD MAPS TS-0011, " HD Maps data contents and formats standard," v1.1, 2020. https://gps.moi.gov.tw/sscenter/introduce/BulletinPage_Info.aspx?ID=101
He, K., Gkioxari, G., Dollár, P., & Girshick, R. (2017). Mask r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 2961-2969).
Landa, J., & Prochazka, D. (2014). Automatic road inventory using LiDAR. Procedia economics and finance, 12, 363-370.
Li, D., & Elberink, S. O. (2013). Optimizing detection of road furniture (pole-like objects) in mobile laser scanner data. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci, 1, 163-168.
Li, Y., Wang, W., Tang, S., Li, D., Wang, Y., Yuan, Z., ... & Xiu, W. (2019). Localization and extraction of road poles in urban areas from mobile laser scanning data. Remote Sensing, 11(4), 401.
Kumar, Chellapilla (2018). Rethinking Maps for Self-Driving. Medium: https://medium.com/lyftlevel5/https-medium-com-lyftlevel5-rethinking-maps-for-self-driving-a147c24758d6
PCL, 2020,” Euclidean Cluster Extraction”: https://pcl.readthedocs.io/projects/tutorials/en/latest/cluster_extraction.html?highlight=euclidean
PCL, 2020,” How to use Random Sample Consensus model”: https://pcl.readthedocs.io/projects/tutorials/en/latest/random_sample_consensus.html#random-sample-consensus
Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. In Advances in neural information processing systems (pp. 91-99).
Schnabel, R., Wahl, R., & Klein, R. (2007, June). Efficient RANSAC for point‐cloud shape detection. In Computer graphics forum (Vol. 26, No. 2, pp. 214-226). Oxford, UK: Blackwell Publishing Ltd.
TAICS TR-0010, "HD Maps Operation Guidelines," v2, 2019. https://gps.moi.gov.tw/sscenter/introduce/BulletinPage_Info.aspx?ID=102
TAICS TR-0016, "Verification and validation guidelines for HD Maps," v0.4, 2020. https://gps.moi.gov.tw/sscenter/introduce/BulletinPage_Info.aspx?ID=100
Tan, M., Wang, B., Wu, Z., Wang, J., & Pan, G. (2016). Weakly supervised metric learning for traffic sign recognition in a LIDAR-equipped vehicle. IEEE Transactions on Intelligent Transportation Systems, 17(5), 1415-1427.
Vo, A. V., Truong-Hong, L., Laefer, D. F., & Bertolotto, M. (2015). Octree-based region growing for point cloud segmentation. ISPRS Journal of Photogrammetry and Remote Sensing, 104, 88-100.
Zhang, K., Chen, S. C., Whitman, D., Shyu, M. L., Yan, J., & Zhang, C. (2003). A progressive morphological filter for removing nonground measurements from airborne LIDAR data. IEEE transactions on geoscience and remote sensing, 41(4), 872-882.
Zhao, H., Xi, X., Wang, C., & Pan, F. (2019). Ground Surface Recognition at Voxel Scale From Mobile Laser Scanning Data in Urban Environment. IEEE Geoscience and Remote Sensing Letters.
Zhou, L., & Deng, Z. (2014, October). LIDAR and vision-based real-time traffic sign detection and recognition algorithm for intelligent vehicle. In 17th international IEEE conference on intelligent transportation systems (ITSC) (pp. 578-583). IEEE.
Zhu, Z., Liang, D., Zhang, S., Huang, X., Li, B., & Hu, S. (2016). Traffic-sign detection and classification in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2110-2118).
張桂華(民109)。Semi-automatic Image Annotation Method for Traffic Sign in Taiwan。國立成功大學測量及空間資訊研究所碩士論文,未出版,中華民國。
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2020-08-07起公開。
  • 同意授權校外瀏覽/列印電子全文服務,於2020-08-07起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw