A Study on Nighttime Brake Light Detection in Low-Exposure Video

• Main Authors: LUO,HAN-DE, 羅瀚德
• Other Authors: Yong-Huai-Houng
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/5cm9hw

碩士 === 景文科技大學 === 電腦與通訊研究所 === 101 === With the development of technology and economics, the number of automobiles has been increasing rapidly, which leads to a higher percentage of car accidents. In order to reduce the rate of traffic accidents, how to develop intelligent driving assisted system has become an important issue. In this study, an effective system, based on computer vision and image processing technology, is proposed in order to detect the nighttime brake light. In order to localize taillight effectively, a low-exposure method determines candidate regions of taillight. With the feature of HSV color space, the accuratetaillight is selected from candidate regions. Finally, this system sends a warning signal to drivers when brakes are detected according to the variation of taillight within continuous image frames.