基於行車影像之危險駕駛偵測系統

Abstract

隨著無線通訊網路技術的進步,加上電子元件價格漸漸的下降以及行車安 全問題越來越受重視,各國對智慧型交通運輸系統(Intelligent Transport Systems, ITS)的研究[1-5]。ITS 利用先進的電子、通信、資訊與感測等技術,以整合人、 路、車的管理策略,其主要的目地為提供即時(real-time)的資訊,使有限運輸資源 發揮最大效用的系統,並增進運輸系統的安全、效率及舒適性,同時也減少交通對 環境的衝擊。 本篇論文對目前發展快速的智慧型影像裝置(如行車紀錄器或具有影像擷取 功能的智慧型手機)，收集視訊資料,經過影像辨識處理過的各項行車特徵資料 (如車道分隔線、車輛種類、車輛速度與車間距離等)。進而開發一套危險駕駛偵 測系統。

Due to the advances on wireless technologies and the dramatic cost down on electronic components, the development of Intelligent Transportation System (ITS) has drawn intensive attention in recent years. ITS applies advanced technologies, such as electronic, communication and sensing technologies, to achieve a safe, efficient, and comfort transportation system. It can also reduce the traffic accidents. In this thesis, we apply an on-broad video device (e.g., driving recorder or smart phone with video component) to collect the traffic information. Then, by pattern recognition techniques, we extract the useful road and traffic features (e.g. Barrier line, vehicle type, vehicle speed and vehicle distance) to develop a Reckless Driving Detection System. The performance evaluation is also given to show the effectiveness of the proposed system.