丁肇隆臺灣大學：工程科學及海洋工程學研究所黃莫凱Huang, Mo-KaiMo-KaiHuang2007-11-262018-06-282007-11-262018-06-282006http://ntur.lib.ntu.edu.tw//handle/246246/51003車輛主動式安全系統是屬於政府重點發展計畫ITS(Intelligent Transport System智慧型運輸系統)中極為重要之一環，利用感測器協助駕駛者感官功能之不足，提高自動控制之程度，以彌補駕駛者因判斷錯誤或技術不足所造成的疏失，減少危險及意外事故之發生。本系統係以影像視覺為基礎，從影像中辨識出車道標線位置及正前方車輛位置等資訊以提供駕駛行車安全資訊。首先路面影像經由車內照後鏡下方之攝影機紀錄，將影像中欲處理的區域取出後，利用Sobel邊緣偵測，將物體邊緣分辨出來，再以車道線向量法找出最可能之車道線位置。結果顯示，本研究所提之方法，在大部分之路況下能準確地辨識出車道標線。其次偵測正前方車輛距離，由於白天和夜晚亮度不同，分別有不同的處理方法，白天以Sobel邊緣偵測，找尋車輛底部陰影位置；晚上則利用車輛尾燈特徵，以估計車輛之位置。最後將所得之資訊提供車道偏離警示及前車追撞警示系統，作為判斷車輛是否行使於車道之安全範圍內。Vehicle active safety system is very important and intelligent transport system is an important developing subject of our government. Different sensing systems have been developed to assist human driving and to avoid the occurrences of dangers or accidents. Our system is vision-based. The images are acquired by a video camera and processed to recognize the position of the lane markers and the relative distance to the front vehicle. A video camera was mounted on a vehicle to catch the sequence of the roadway. From each recorded image, a region of interest was decided. In the region of interest, the edges of objects were detected using Sobel edge detection method. Then a lane-vector method was used to find the most possible positions of the lane markers. Results showed that our method can recognize the positions of lane markers correctly at different road conditions. In front-vehicle detection two methods were used in daytime and at night. In the daytime, we detect the shadow of the front vehicle on the ground to calculate the relative distance; however, during the night the paired tail lights of cars were used to estimate the position of the front vehicle. Those information from the past image processing were used to provide lane-departure warning and collision-warning systems to judge if the car is driving safely.中文摘要 Ⅰ 英文摘要 Ⅱ 目錄 Ⅲ 表目錄 Ⅴ 圖目錄 Ⅵ 符號說明 Ⅸ 第一章 前言 1 1.1 研究動機與目的...........................................1 1.2 相關文獻探討………………………………………4 1.3 論文架構…………………………………………..10 第二章 系統架構 11 2.1 設備與裝置................................................11 2.2 基本假設…………………………………………..12 2.3 相機校正…………………………………………..12 2.4 影像處理流程…………………………………….14 2.4.1 影像擷取…………………………………….15 2.4.2 影像處理…………………………………….16 2.4.3 車道線及前車距離偵測…………………..17 2.4.4 逆透視轉換………………………………….18 2.4.5 警示發佈系統………………………………21 第三章 道路及前車偵測 23 3.1 邊界偵測…………………………………………23 3.1.1 車道標線之特徵……………………………23 3.1.2 邊界偵測介紹………………………………24 3.1.3 道路標線識別………………………………32 3.2 前方車輛的偵測………………………………….38 3.2.1 車輛在影像上之特徵……………………..38 3.2.2 前車偵測之方法介紹……………………..39 3.3 速度計算…………………………………………..55 3.3.1 側向速度計算之方法介紹………………..55 3.3.2 相對車速計算之方法介紹………………..57 3.4 警示系統…………………………………………..58 3.4.1 車道偏離警示系統…………………………58 3.4.2 前車追撞警示系統…………………………59 3.5 即時處理…………………………………………..60 第四章 實驗結果 61 4.1 相機校正…………………………………………..61 4.2 影像擷取及處理………………………………….62 4.3 車道標線偵測結果……………………………….66 4.4 車道偏離警示系統……………………………….70 4.5 前方車輛偵測結果……………………………….72 4.6 前方車輛追撞警示系統…………………………75 4.7 IPM逆透視轉換之驗證………………………….78 第五章 結論與未來目標 82 參考文獻 852665327 bytesapplication/pdfen-US車道線偵測前方車輛偵測lane detectionvehicle detectionthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/51003/1/ntu-95-R93525016-1.pdf