傅立成Fu, Li-Chen臺灣大學:資訊工程學研究所邱一航Chiu, Yi-HangYi-HangChiu2010-05-182018-07-052010-05-182018-07-052008U0001-2307200817205600http://ntur.lib.ntu.edu.tw//handle/246246/183612每年都有許多人因為交通意外而受傷或死亡。因此，本篇論文希望藉由影像視覺，偵測後方危險的車輛，以警告和保護駕駛者。這些危險的車輛，在不同的交通環境或天氣環境下的不同種類車輛，或因為相機的視角，造成後方車輛影像呈現部分可見或完全可見車輛的情況下，都必須偵測得到。本篇論文所提出的方法，以電腦視覺的方式，將車輛以一組車輛局部影像來描述，並且整合和選擇具鑑別性及差異性的車輛局部影像特徵，來進行後方車輛偵測。其中基於修改AdaBoost演算法，來整合差異性的車輛局部影像特徵，並找出具鑑別性的局部影像特徵。其演算法包含多個弱分類器(Weak Classifier)，弱分類器包含弱特徵描述分類器，對於描述局部車輛外觀的特徵之鑑別度做分類，以及弱特徵空間分類器，對於車輛局部區域與車輛中心的幾何關係之穩定度做分類。其中弱特徵描述分類器用來評估不同局部車輛特徵的外觀之鍵別性，以及特徵空間分類器來評估不同車輛局部特徵對於車輛中心的幾何關係之穩定度。由經修改過後的AdaBoost演算法來選擇不同車輛局部特徵。在影像空間下偵測車輛時，這些經學習與選擇的具差異性且具鑑別性的車輛局部區域，對於可能車輛的中心位置作投票，當圖像中為多數票數的區域即為車輛中心的位置。Every year many people are injured and killed in traffic accidents. In order to warn and protect the driver in the potential accidents, we develop a vision system to detect whatever style rear vehicles under various traffic or weather conditions. Otherwise, in the vision system, the incomplete rear vehicle is shown in the image due to the limitation of field of view of the camera. We propose a vision system detecting rear vehicles, whether occluded or not. The system selects and integrates different parts of the rear vehicle pattern; these parts are represented by complementary characteristics. There are two kinds of classifiers for the part selection. First, the appearances of each vehicle’s part with different feature types are evaluated by the representative classifiers. Second, the geometry of the parts is modeled as a spatial classifier; each spatial classifier evaluates the stability of vehicle location estimation of each part. Representative classifiers and the spatial classifiers are learnt and integrated according to the performance of each part, including the discriminabilities of appearance and the stabilities of the geometry estimation. The modified error rate function of a boosting algorithm selects parts of the rear vehicle with complementary discriminative features. These selected parts vote the possible vehicle’s location in the image; the majority votes locate the possible rear vehicles.誌謝 i文摘要 iiBSTRACT iiiONTENTS ivIST OF FIGURES viiIST OF TABLES xhapter 1 Introduction 1.1 Motivation 1.2 Challenges 4.3 Related Work 4.4 Objective 8.5 System Overview 9.6 Thesis Organization 11hapter 2 Preliminary 12.1 Recognition Issue 12.1.1 Feature Representation 12.1.2 Training Model 13.2 Detection Issue 17.2.1 Random Approach 17.2.2 Feature-based Approach 17.2.3 Model-based Approach 18hapter 3 Selection in Parts of the Rear Vehicle 23.1 Part Representation and Evaluation 23.1.1 Knowledge-based Feature 24.1.2 Region-based Feature 25.2 Building Discriminative Pools 27.3 Selecting Discriminative Parts 33.3.1 Weak Representative Classifiers 35.3.2 Weak Spatial Classifiers 36.4 Discriminative Part Selection 40.5 Analysis of the Training Error Rate 45hapter 4 Rear Vehicle Detection and Warning 47.1 Rear Vehicle Detection 47.1.1 Part Detection using Weak Representative Classifiers 47.1.2 Confidence of an Image 49.1.3 Vehicle Detection with Mean-Shift Algorithm 50.2 Warning System in Rear Traffic Scenes 51.2.1 Find Rear Vehicles in Regions of Interest 51.2.2 Collision Avoidance Mechanism 52hapter 5 Experiments 55.1 Environment Setup 55.1.1 Hardware Description 55.1.2 Environment Description 56.2 Training Parameters 56.3 Detection Results 56.4 Performance Analysis and Discussion 58hapter 6 Conclusion 60EFERENCE 62application/pdf731747 bytesapplication/pdfen-US車輛偵測後方車輛偵測物體局部偵測特徵選擇特徵整合Vehicle DetectionRear Vehicle DetectionPart-based Object DetectionFeature Selection, Feature Integration[SDGs]SDG3thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/183612/1/ntu-97-R95922063-1.pdf