在監控交通與治安的系統中，車輛經常是最重要的監控目標之一。本論文提出以顏色為依據的車輛自動分類方法。將檢測到的車輛影像經由二階層分類器進行分類。車輛被分為彩色車輛和灰階車輛。彩色車輛和灰階車輛將分別處理。我們研究方法著重於分類器設計和彩色車輛的特徵擷取。特徵提取的核心概念是將車輛分為階層的樹狀結構，將主體與車輪，車窗，和其他汽車零件分離。我們的方法是將只含有車子主體的圖像用於彩色分類架構。實驗結果表明，本文所提的車輛顏色辨識方法兼具效率與效能,可應用於即時監控系統。

Vehicles are one of the main detection targets of the traffic and security video surveillance system. In this thesis, we propose an automatic vehicle color identification method for vehicle classification. The detected vehicle goes through a two layer classifier. Then, vehicle is classified into chromatic vehicle and achromatic vehicle. We focus on the classifier design and chromatic vehicle’s feature. The main idea of the feature extraction scheme is to divide a vehicle into a hierarchical coarse-to-fine structure to extract its wheels, windows, main body, and other auto parts. In the proposed method, the main body alone is classified by based on a chromatic scheme. Experimental results show that the proposed scheme is efficient and effective and the proposed vehicle color identification is suitable for real-time surveillance applications.

Contents
中文審定書………………………………………………………………………………i
英文審定書………………………………………………………….…………………..ii
致謝……………………………………………………………………………………..iii
中文摘要………………………………………………………………………………... v
Abstract………………………………………………………………………………….vi
Contents………………………………...……………………………………………… vii
List of Figures………………………………………………………………………... viii
List of Tables…………………………………………………………………………. ix
Chapter 1 Introduction……………………………………………………………………………... 1
1.1 Overview of Vehicle Color Classification………………………………….1
1.2 Motivation…………………………………………………………………. 4
1.3 Contribution………………………………………………………………... 5
1.4 Organization……………………………………………………………….. 6
Chapter 2 Background Review…………………………………………………………………….. 7
2.1 Gaussian Mixture Model…………………………………………………... 8
2.2 Support Vector Machine…………………………………………………..12
2.3 Expectation-Maximization Algorithm…………………………………….16
Chapter 3 Proposed Feature Extraction And Classification………………………….…………… 20
3.1 Feature Extracting Process……………………………………………….. 23
3.2 Recursively Algorithm for Extracting the Representative Color……….. 26
3.3 Selection of the Representative Image…………………………………….31
3.4 Achromatic Classifier…………………………………………………….. 35
3.5 Chromatic Classifier……………………………………………………… 37
3.5.1 Color Model Update Mechanism…………………………………..40
3.5.2 Classification Process……………………………………………...43
Chapter 4 Experimental Results……………..……………………………………………………. 48
Chapter 5 Conclusion……………………………………………………………………………... 53
Reference………………………………………………………………………………. 55

List of Figures
Figure 2-1. Example of OSH ...................................................................................... 13
Figure 2-2. Non-linearly separable example............................................................... 14
Figure 2-3. Example of modified maximum margin method ..................................... 15
Figure 2-4. Flow char of EM algorithm ...................................................................... 19
Figure 3-1. Flow char of propose scheme .................................................................. 21
Figure 3-2. Example of saturation image ................................................................... 24
Figure 3-3. Example of grayscale image ................................................................... 25
Figure 3-4. partition result of the detected vehicle ..................................................... 30
Figure 3-5. The example of color difference .............................................................. 32
Figure 3-6. The example of hue feature extraction ..................................................... 34
Figure 3-7. Structure of two layer classifier ............................................................... 36
Figure 3-8. Flow char of chromatic classifier ............................................................. 38
Figure 4-1. Detected original vehicle images. ............................................................ 49
Figure 4-2. Representative images. ............................................................................. 50
Figure 4-3. Classification results ................................................................................ 52

List of Tables
Table 4-1. Training detail of first layer classifier and achromatic classifier ............... 51
Table 4-2. Accuracy of chromatic classification ......................................................... 52

[1] S. M. Lee, J. H. Xin, S. Westland, “Evaluation of image similarity by histogram intersection,” Color Research and Application, Vol. 30, No. 4, pp. 265-274, 2005.
[2] O. Chapelle, P. Haffner, V. N. Vapnik, “Support vector machines for histogram-based image classification,” IEEE Transactions on Neural Networks, Vol. 10, No. 5, pp.1055-1064, 2004.
[3] Sural, G. Qian, S. Pramanik, “Segmentation and histogram generation using the HSV color space for image retrieval,” Proc. of IEEE International Conference on Image Processing, Vol.2, pp.589-592, 2002.
[4] Wu, Yi-Ta, Jau-Hong Kao and Ming-Yu Shih, “A vehicle color classification method for video surveillance system concerning model-based background subtraction," Proc. of of the 11th Pacific Rim Conference on Advances in Multimedia Information Processing, pp. 369-380, 2010
[5] Kim, Ku-Jin, Sun-Mi Park and Yoo-Joo Choi, “Deciding the number of color histogram bins for vehicle color recognition," Proc. of Asia-Pacific Services Computing Conference, pp. 134-138, 2008.
[6] Ye Li, Bo Li, Bin Tian, Qingming Yao, “Vehicle detection based on the AND–OR graph for congested traffic conditions," IEEE Transactions on Intelligent Transportation Systems, Vol. 14, No. 5, pp.984-993, 2013
[7] C. Stauffer and W. E. L. Grimson, “Learning Patterns of Activity Using Real-Time Tracking,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 747-757, 2000.
[8] Cortes, C. and Vapnik, V., “Support-vector networks,” Machine Learning, 20, pp. 273-279, 1995
[9] Boser, B. E., Guyon, I. M., Vapnik, V. N., “A training algorithm for optimal margin classifiers,” Proceedings of the fifth annual workshop on Computational learning theory, pp. 144-152, 1992
[10] J. L. W. V. Jensen, “ Sur les fonctions convexes et les inégalités entre les valeurs moyennes,” Acta Mathematica, Vol. 30, Issue. 1, pp 175-193, 1906
[11] Chih-Yang Lin, Cheng-Hao Yeh, Chia-Hung Yeh, “Real-time Vehicle Color Identification for Surveillance Videos,” Conference on Electronics, Communications and Computers (CONIELECOMP), pp 59 - 64, 2014