現今車輛自動駕駛主要利用RADAR、LiDAR及電腦視覺對車輛周圍環境進行感測與識別以執行保距、跟隨及迴避等動作，定位及行駛則使用衛星定位系統作為其導航方式。衛星定位系統具有定位精度高、移動定位及近乎全球覆蓋等優點，但同時也具備缺點，如行車時周圍天空有遮蔽物或是行駛於山區時，可能因訊號不佳導致定位精準度受到影響，或是行駛於城市街道區接獲轉彎提示時，用路人正好面臨許多巷弄等小路，無法將GPS的導引與現實情況正確結合。電腦視覺具有影像導航的潛力，但現今技術尚無法支援其獨立達成，然其可在某些特定環境下達到輔助導航的目標。影像輔助系統與衛星系統共同運作將能提高精準度，降低GPS訊號不佳對定位造成的影響，亦可提升GPS在路況複雜區域面對轉彎時的精準度。
本文提出一個利用影像處理及類光流為基礎的導航資訊預處理系統，藉由加速穩健特徵 (Speed Up Robust Features, SURF)算法對預先錄製的行車畫面或是連續街景圖進行影像處理，提取關鍵畫格組。藉由本文提出之航行資訊自動判定系統賦予關鍵畫格組相對之導航資訊，建構具有導航資訊的影像數據集作為影像導航依據。

The aim of this research is to establish an automatic image pre-processing and navigation information judgment system to produce image datasets that can support computer vision navigation. The road image library is collecting from driving recorders. Feature points of the road images were extracted by the SURF algorithm and feature point matching was performed. If the match result is greater than the threshold it means that a new key frame has been found, if it is below the threshold then the image is the subsidiary frame belonging to the key frame.
Each key frame and its subsidiary frames form a key frame group, and then the displacement of the matching feature points between the two frames is used to obtain the feature motion vectors. The distribution data which is calculated from the angles calculated from the feature motion vectors, is judged by the navigation judgement formula to determine the navigation information. The navigation information and key frame groups are integrated into the image navigation datasets.
The datasets is converted into the segment database and corrections are performed. Feedback from the segment database then helps to improve the image navigation datasets. In this paper, the system processes general road images to output navigation information, and its accuracy can reach almost 99%. However, the navigation information is not accurate enough for specific roads. This research will further improve the accuracy of the image navigation datasets so that it can be used as a database for computer vision navigation in the future.

Bay, H., Tuytelaars, T. and Van Gool, L. (2006). SURF: Speeded Up Robust Features. Computer Vision and Image Understanding - CVIU. 110. 404-417.
Bay, H., Ess, A., T. Tuytelaars, Van Gool, L. (2008). Speeded-Up Robust Features (SURF). Computer Vision and Image Understanding, Volume 110, Issue 3, Pages 346-359.
Durrant-Whyte, H., Bailey, T. (2006). Simultaneous localization and mapping: part I. IEEE Robotics & Automation Magazine, vol. 13, no. 2, pp. 99-110, June 2006, doi:10.1109/MRA.2006.1638022.
Horn, B.K.P., Brian, G. (1981). Schunck. Determining optical flow. Artificial Intelligence, Volume 17, Issues 1–3, Pages 185-203, ISSN 0004-3702.
Lowe, D.G. (1999). Object recognition from local scale-invariant features. Proceedings of the Seventh IEEE International Conference on Computer Vision, pp. 1150-1157 vol.2, doi: 10.1109/ICCV.1999.790410.
Lowe, D.G. (2001). Object Recognition from Local Scale-Invariant Features. Proceedings of the IEEE International Conference on Computer Vision. 2.
Lowe, D.G. (2004). Distinctive Image Features from Scale-Invariant Keypoints. International Journal of Computer Vision 60, 91–110.
Mistry, D., Banerjee, A. (2017). Comparison of Feature Detection and Matching Approaches: SIFT and SURF. GRD Journals- Global Research and Development Journal for Engineering. 2. 7-13.
Pan, Y., Sun , X. and Wu, F. (2020). Enriching Optical Flow with Appearance Information for Action Recognition. 2020 IEEE International Conference on Visual Communications and Image Processing (VCIP), Macau, China, 2020, pp. 251-254, doi:10.1109/VCIP49819.2020.9301827.
Raket, L.L. (2012). Local smoothness for global optical flow. 2012 19th IEEE International Conference on Image Processing, Orlando, FL, USA, 2012, pp. 1-4, doi:10.1109/ICIP.2012.6674231.
Rousseeuw, P. J., Croux, C. (1993). Alternatives to the Median Absolute Deviation. Journal of the American Statistical Association, 88(424), 1273–1283.
Ziyun, L., Wei, L. (2010). The Compensated HS Optical Flow Estimation Based on Matching Harris Corner Points. 2010 International Conference on Electrical and Control Engineering, Wuhan, China, 2010, pp. 2279-2282, doi: 10.1109/iCECE.2010.562.
Zuiderveld, K. (1994). Contrast Limited Adaptive Histograph Equalization. Graphic Gems IV. San Diego: Academic Press Professional. 474–485.
冷玉琦（2009）。機械視覺於道路駕駛偏航檢知系統之研究。碩士論文。國立成功大學。