智慧型運輸系統(Intelligent Transportation System)在世界各地已發展多年。其目的在於藉由先進技術之輔助以改善與加強地面交通之安全與效率。利用不同之基礎建設與感測器科技，智慧型運輸系統可以增進交通運輸之效率，減少交通事故，進一步並能減少能源之浪費。
自動汽車駕駛系統是智慧型運輸系統的一部分，其目標在減少因駕駛人的疏忽所造成之意外事故。而障礙物偵測與警告系統在這類研究中扮演了一個重要的角色。
在這方面的相關研究中，各種不同類型的感測器被使用，其中包括主動式感測器(Active Sensor)與被動式感測器(Passive Sensor)。在本論文所提出的系統中，我們選擇以電腦視覺作為感測的方式，相對於其他感測科技，電腦視覺有較大的偵測範圍並可提供豐富的資訊。
我們定義了三種不同的障礙物種類，分別是: 行人，汽車及其他障礙物。而由於相機本身具有會移動的特性(Ego-Motion)，從而增加了辨識的困難度。在本論文中，針對不同種類障礙物之特性，我們利用不同的方式來作辨識。
對於行人的辨識，我們使用立體視覺來找出行人可能存在之位置，並提出了一套模板比對演算法來進行最後的確認。在所提出的演算法中，我們使用基因演算法(Genetic Algorithm)來對資料作叢集(Data Clustering)的動作以有效減少比對的次數並增進辨識之效率。而對於汽車，我們利用汽車底部的陰影來擷取汽車可能的位置，並利用汽車外型之特性來作最後的確認。另外為增進系統之效率，當物體被偵測到之後，我們會利用追蹤(Tracking)來預測下次障礙物之位置，並檢查區域的顏色分布以確認物體是否存在。
　這套系統不只適用於單純的環境如高速公路，對於具有複雜背景的環境如市區道路同樣適用。因此，這套系統可以同時增進汽車駕駛人與市區行人之安全。

Intelligent Transportation System (ITS) has been studied for many years all over the world. The goal of such system is to apply advanced technologies to improve the safety and efficiency of surface transportation system. By cooperating with different kinds of infrastructures and sensor technologies, ITS can help to improve the traffic efficiency, reduce the accidents, and decrease the consumption of energy.
Autonomous driving system is one part of the ITS. Such a system can assist to prevent the traffic accidents caused by the negligence of the driver. Obstacle detection and warning mechanism plays an important role in this research field.
Active sensors and passive sensors are used to achieve this goal in many researches. In the proposed system, we adopt the computer vision technology because of its large detecting range and abundant information when compared with other kinds of sensors.
Three categories of obstacles are defined in the proposed system: Pedestrian, Vehicle and Others. Due to different characteristics of these obstacles, we adopt different methods for detection of different kinds of obstacles.
For detection of pedestrian, the potential pedestrian regions are extracted via a simplified fast stereovision method. A template database composed of different gaits is reorganized by adopting the Genetic K-Means Algorithm (GKA) for the verification of pedestrian. The potential regions are compared only with the representative templates from different clusters to improve the efficiency and the M-estimation Hausdorff Distance is used as the metric. On the other hand, the potential locations of vehicle are extracted by the search of their dark underneath. Symmetry properties and edge ratios of these regions are then checked for verification.
After we obtain the locations of these obstacles, their color histogram and geometric positions are recorded for future tracking. A simple extrapolation method is used to predict the next locations of the tracked obstacles. These predicted locations are verified by the concept of intersection of color histograms.
The proposed system is suitable for both the simplified environment such as freeway and the urban environment with complex background. Thus, the result here improves traffic safety not only for drivers, but also for all pedestrians on the road.

CHAPTER 1 INTRODUCTION 1
1.1 MOTIVATION 1
1.2 BRIEF SURVEY OF RELATED RESEARCHES 3
1.3 OBJECTIVE 5
1.4 ORGANIZATION 5
CHAPTER 2 PRELIMINARY KNOWLEDGE 7
2.1 PROBLEM DEFINITION 7
2.1.1 Existence Problem 8
2.1.2 Verification Problem 9
2.1.3 Location Problem 10
2.1.4 Safety Problem 12
2.2 CAMERA CONFIGURATION 15
2.3 SYSTEM DESCRIPTION 16
CHAPTER 3 DETECTION OF PEDESTRIAN 18
3.1 OVERVIEW 18
3.2 PEDESTRIAN DETECTION PROCEDURE 20
3.3 POTENTIAL PEDESTRIAN REGION 20
3.3.1 Stereo Vision 22
3.3.2 Fast Binocular Correspondence 22
3.3.3 Sparse Depth Map 25
3.3.4 Disparity Histogram 26
3.3.5 Windows Fitting 27
3.3.6 Summary 30
3.4 VERIFICATION OF PEDESTRIAN 31
3.4.1 Hausdorff Distance 33
3.4.2 Data Clustering 40
3.4.3 Template Matching Based Verification 45
3.4.4 Summary 47
CHAPTER 4 DETECTION OF VEHICLE 50
4.1 OVERVIEW 50
4.2 VEHICLE DETECTION PROCEDURE 51
4.3 POTENTIAL VEHICLE POSITION 53
4.3.1 Sign Pattern 53
4.3.2 Horizontal Line Filtering 56
4.3.3 Positional Filtering 57
4.3.4 Summary 57
4.4 VERIFICATION OF VEHICLE 60
4.4.1 Symmetry and Edge Filtering 60
4.4.2 Post Filtering 61
4.4.3 Summary 63
CHAPTER 5 ENHANCING PERFORMANCE BY TRACKING 65
5.1 OVERVIEW 65
5.2 POSITION PREDICTION 66
5.3 POSITION VERIFICATION 67
5.3.1 Color Indexing 67
5.3.2 Verification Using Color Indexing 68
5.4 TRACKING PROCEDURE 69
5.4.1 Information Combination Phase 70
5.4.2 Position Prediction Phase 71
CHAPTER 6 EXPERIMENT 74
6.1 ENVIRONMENT DESCRIPTION 74
6.2 EXPERIMENT RESULTS 78
6.2.1 Pedestrian Detection 78
6.2.2 Vehicle Detection 90
6.2.3 Tracking 95
CHAPTER 7 CONCLUSION 96
REFERENCE 99

[1] M. Bertozzi and A. Broggi, "GOLD: a parallel real-time stereo vision system for generic obstacle and lane detection," IEEE Transactions on Image Processing, Vol. 7, No 1, pp. 62-81, Jan. 1998.
[2] U. Franke and I. Kutzbach, "Fast stereo based object detection for Stop&Go traffic," Proceedings of the IEEE Intelligent Vehicles Symposium, pp. 339-344, 1996.
[3] N. D. Matthews, P. E. An, and C. J. Harris, "Vehicle detection and recognition for autonomous intelligent cruise control," Image, Speech and Intelligent Systems, 1995/6 Research Journal of Dept. of Electronics and Computer Science, University of Southampton, UK, 1995.
[4] M. Bertozzi, A. Broggi, A. Fascioli, and S. Nichele, "Stereo Vision-based Vehicle Detection," Proceedings of the IEEE Intelligent Vehicles Symposium 2000, pp. 39-44, Oct. 2000.
[5] M. Betke, E. Haritaoglu, and L. S. Davis, Multiple vehicle detection and tracking in hard real time, Technical Report CS-TR-3667, University of Maryland, College Park, July 1996.
[6] Q.-T. Luong, J. Weber, D. Koller, and J. Malik. "An integrated stereo-based approach to automatic vehicle guidance," Proceedings of the International Conference on Computer Vision, pp. 52-57, Cambridge, MA, 1995.
[7] A. Broggi, M. Bertozzi, A. Fascioli, and M. Sechi, "Shape-based pedestrian detection," Proceedings of the IEEE Intelligent Vehicles Symposium, pp. 215-220, Oct. 2000
[8] C. Papageorgiou and T. Poggio, "Trainable Pedestrian Detection," Proceedings of International Conference on Image Processing, Vol. 4, pp. 35-39, 1999.
[9] A. Mohan, C. Papageorgiou, and T. Poggio, "Example-based object detection in images by components," IEEE Transactions on Pattern Analysis and Machine Intelligence, volume 23, pages 349--361, April 2001.
[10] D.M. Gavrila and V. Philomin, "Real-Time Object Detection For "SMART" Vehicles," In Proceedings of IEEE International Conference on Computer Vision, Vol. 1, pp. 87-93, 1999.
[11] C. Curio, J. Edelbrunner, T. Kalinke, C. Tzomakas, and W. V. Seelen, "Walking Pedestrian Recognition," IEEE Transactions on intelligent transportation system, Vol. 1, No. 3, Sept. 2000.
[12] http://www.ets.uidaho.edu/niatt_labmanual/index.htm Transportation Engineering On Line Manual.
[13] C. Y. Liu, "Computer Vision Based Movable Obstacle Detection and Recognition for Vehicle Driving," Master thesis, Dept. of Computer Science and Information Engineering, National Taiwan University, 2000.
[14] Z. T. Sun, "On Road Computer-Vision Based Obstacle Detection," Master thesis, Dept. of Computer Science and Information Engineering, National Taiwan University, 2001.
[15] A. Broggi, M. Bertozzi, A. Fascioli, C. G. L. Bianco, and A. Piazzi, "Visual Perception of Obstacles and Vehicles for Platooning," IEEE Transactions on intelligent transportation system, Vol. 1, No. 3, pp. 164-176, 2000.
[16] W. Rucklidge, Efficient Visual Recognition Using the Hausdorff Distance, Springer-Verlag, 1996.
[17] D. P. Huttenlocher, G. A. Klanderman, and W. J. Rucklidge, "Comparing images using the Hausdorff distance," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 15, No. 9, pp. 850-863, Sept. 1993.
[18] R. Azencott, F. Durbin, and J. Paumard, "Multiscale identification of building in compressed large aerial scenes," Proceedings of International Conf. Pattern Recognition, Vol. 2, pp. 974-978, Aug. 1996.
[19] M. P. Dubuisson and A. K. Jain, "A modified Hausdorff distance for object matching," Proceedings of International Conf. Pattern Recognition, pp. 566-568, Oct. 1994
[20] D. G. Sim, O. K. Kwon, and R. H. Park, "Object matching algorithms using robust Hausdorff distance measures," IEEE Transactions on Image Processing, Vol. 8, No. 3, pp. 425-429, March 1999
[21] Y. Dai Y. Nakano, "Face-texture model-based on sgld and its application in face detection in a color scene," Pattern Recognition, Vol. 29, pp. 1007-1017, June 1996.
[22] H. Mori, N. Moghadam, and T. Matsushita, "On-Line Vehicle and Pedestrian Detections Based on Sign Pattern," IEEE Transactions on industrial electronics, Vol. 41, No. 4, Aug. 1994.
[23] M. Swain and D. Ballard, "Color indexing," International Journal of Computer Vision, Vol. 7, pp. 11-32, 1991.
[24] M. P. Dubuisson and A.E. Jain, ŖD matching of 3D moving objects in color outdoor scenes," Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 887-891, 1994.
[25] J. H. Holland. Adaptation in natural and artificial systems. The University of Michigan Press, Ann Arbor, 1975.
[26] K. Krishna and M. N. Murty, "Genetic K-Means Algorithm," IEEE Transactions on Systems, Man and Cybernetics, Vol. 29, pp. 433-439, June 1999.