本論文提出一套基於統計原理的車輛顏色辨識技術，可對行進中車輛的影像進行其與背景間的關聯分析，並依不同方向車輛的車輛特徵進行分類和辨識。基於車輛顏色不容易隱藏的特性，以車輛顏色作為辨識特徵更能增加監控系統架設的自由度。因此，本研究的主要目的在於提供車輛顏色辨識資訊以增加交通監控之影像分析處理系統更多樣的資訊從而達到節省人力和增加車輛辨識的可靠度。
　　在擷取交通資訊的過程裡，我們使用網路攝影機拍攝的車流影片中所擷取下來的車輛影像，依其HSV（Hue、Saturation、Value）值，在色彩空間中設定不同的分類門檻值對待測影像中像素的顏色資訊進行多層式的篩選，分類並統計其比例，再利用本論文提出的關聯式分析模型從分類結果所得到的比例分佈數據判斷車輛的顏色。在本方法中，我們使用比值去了解影像中物件與背景間的關聯性，降低物件被背景亮度的影響，使其具有對外在光源影響的容忍力。另外也依據車輛不同的行進方向對門檻值進行對應的調整以提高辨識的正確性。實驗結果證明，本論文提出的方法對正向、側向以及俯角三個方向所拍攝的車輛非常有效，其辨識車輛顏色的平均正確率達到90.2%。

In this thesis, we proposed a method that can detect the color of a vehicle based on the statistical theory. The method can analyze the relationship between moving vehicles and background and then classify the vehicles according to moving directions. Then recognize the color of a vehicle in different classifications. Owing to the not being hidden color of a vehicle, it can increase the degree of freedom for setting up a surveillance system. Therefore, the main purpose of this thesis is to provide the vehicle color recognition information to enhance the database of a traffic video surveillance system, and then save the manpower and improve the reliability of the recognition system.
In the process of capturing traffic information, we use the pictures capturing by a CCD camera and setting different thresholds in HSV (hue, saturation, and value) color space to classify all the pixels in the images of vehicles in different directions. The test image is going through the multilayer filters to classify to which class it belongs. Then based on the concentration ratio and proportion for each class, it then can recognize the color of a vehicle of the test image. In addition, this method uses the concentration ratio between objects and background to learn the correlation between them, so it also has the ability to improve the light source effect in the outdoor environment. Besides that, we set three different parameters to improve the recognition rates for three different vehicle moving directions. Experimental results show that the proposed method is very robust and efficient for the vehicle in front, side and overlook directions and the average color recognition rates are up to 90.2%.

[1] http://www.motc.gov.tw/mocwebGIP/wSite/mp?mp=1 交通部全球資訊網
[2] 張瓊芳、薛繼光，「錄影中請微笑－無所不在的監視器」，光華雜誌，第三十四卷，第九期，第82頁 (2009)
[3] http://www.cepd.gov.tw/m1.aspx?sNo=0011499 行政院經濟建設委員會
[4] http://news.chinatimes.com/ 中時電子報 1月7日, 2010
[5] Y. Rubner, J. Pucizha, C. Tomasi, and J. Buhmann., “Empirical evaluation of dissmilarity measures for color and texture” , Computer Vision and Image Understanding, pages 25-27 (2001)
[6] Y. Rui, T. Huang, S. Chang, “Image Retrieval: Current Techniques Promising Directions, and Open Issues” , Journal of Visual Communication and Image Representation, 10(4):39-41 (1999)
[7] Miriam Butzke1, Alexandre G. Silva1, Marcelo da S. Hounsell1, Maur′ıcio A. Pillon, “Automatic Recognition of Vehicle Attributes –Color Classification and Logo Segmentation” , Hifen, Uruguaiana, v. 32 – n°62–II Semestre–Ano 2008–ISSN 1983-6511 (2008)
[8] K. Yamaguchi, Y. Nagaya, K. Ueda, H. Nemoto, and M. Nakagawa, “A method for identifying specific vehicles using template matching” , in Proc. IEEE Int. Conf. Intelligent Transportation Systems, pp.8-13 (1999)
[9] P. Davies, N. Emmott, and N. Ayland, ”License plate recognition technology for toll violation enforcement” , Inst. Elect. Eng. Colloquium Image Analysis for Transport Applications, pp. 7/1-7/5 (1990)
[10] R. A. Lotufo, A. D. Morgan, and A. S. Johnson, ”Automatic numberplate recognition” , Inst. Elect. Eng. Colloquium on Image Analysis for Transport Applications, pp. 6/1-6/6, 1990
[11] 高智源, 「基於即時影像處理之交通資訊整合系統」，碩士論文，國立台灣科技大學，台北 (2005)
[12]Xiaohu Lu and Hong Zhang, “Color Classification Using Adaptive Dichromatic Model” , Proceedings of the 2006 IEEE International Conference on Robotics and Automation Orlando, Florida - May (2006)
[13] Alireza Kashanipour,Narges Shamshiri Milani, Amir Reza Kashanipour, Hadi Haji Eghrary, “Robust Color Classification Using Fuzzy Rule-Based Particle Swarm Optimization” , Congress on Image and Signal Processing (2008)
[14] S. D. Buluswar, B. A. Draper, “Color machine vision for autonomous vehicles” , Engineering Applications of Artificial Intelligence, Vol. 11, No. 2, pp. 245-256 (1998)
[15] Ku-Jin Kim1, Sun-Mi Park2, Yoo-Joo Choi3, “Deciding the Number of Color Histogram Bins for Vehicle Color Recognition” , IEEE Asia-Pacific Services Computing Conference (2008)
[16] H. Inoue, L. Mingzhe, S. Kamijo, “Vehicle segmentation by edge classification method and the S-T MRF model,” Proc. of IEEE International Conference on Systems, Man and Cybernetics, Vol. 1, 2006, pp. 370-376 (2006)
[17] J. –W. Lee, I. –S. Kweon, “Vehicle segmentation using evidential reasoning” , Proc. of IEEE/RSJ International Conference on Intelligent Robots and Systems, Vol. 2, pp. 880-885 (1997)
[18] Z. –F. Liu, Z. You, “A real-time vision-based vehicle tracking and traffic surveillance” , Proc. of 8th ACIS International Conference on SNPD 2007, Vol. 1, pp. 174-179 (2007)
[19] Jeong-Woo Son, Seong-Bae Park and Ku-Jin Kim “A Convolution Kernel Method for Color Recognition” , Sixth International Conference on Advance Language Processing and Web Information Technology, 2007
[20] 李志君，「基礎色彩學」，色彩技術實驗室，台北，台灣 (2004)
[21] Cowlishaw, M. F.”Fundamental requirements for picture presentation”, Proc. Society for Information Display，26（2）：101–107 (1985)
[22] http://zh.wikipedia.org/ 維基百科
[23] http://www2.dupont.com/Taiwan_Country_Site/zh_TW/index.html “杜邦2008汽車流行色彩報告,” 1月21日, 2009