A Computer Vision-based System for Acquiring Information of Driving Environment

• Main Authors: Heng-Ka Chan, 陳慶嘉
• Other Authors: Tsung-Ying Sun
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/42967544936755948505

碩士 === 國立東華大學 === 電機工程學系 === 94 === This study is dedicated to the field of Intelligent Transportation System (ITS) for driving safety enhancements. In recent decade, considerable attention and effort has been paid to the field, which helps to meet with the challenges in transportation worldwide. The research of ITS is synthesized of a broad range of diverse technologies including information processing, communications, control and electronics. Joining these advanced technologies is not just to enhance transportation but to considerably save lives, time and money. For safety driving, drivers’ awareness and condition are two major concerns in keeping the host vehicle on the suitable position on a lane and to avoid impending situations, which are aimed to keep safety for both the host vehicle and other road users. The main purpose of this study is to enhance driving safety on real-world driving environment with the above issues, by means of developing an intelligent driving environment acquisition system based on computer visual perception. Inevitable information demanded by safety driving including lane- markings and rear-braking-lights of preceding vehicles are extracted by the proposed system. Novel methodologies for lane-marking and rear-braking-light detection are derived in this study. Algorithms including Fuzzy c-Means (FCM) clustering and the well-known Otsu Method are reformed to achieve the goal of rapid segmentations of lane-markings and rear-braking-lights on standard traffic scenes respectively. The color information is utilized to the most extreme as hybrid color models are used with adaptive mechanism, which enhance the precision of recognition under various driving environments. The robustness of the system is established by the rapid computation and high precision, which is capable for real-time applications for intelligent vehicles.