Omni-Directional Lane Detection and Vehicle Motion Estimation with Multi-Camera Vision

• Main Authors: Tai-Chia Tsao, 曹太嘉
• Other Authors: 張文中
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/vtbhp4

碩士 === 國立臺北科技大學 === 電機工程系研究所 === 96 === Driver assistant system has been an active research topic in intelligent transportation systems. It could be used to provide surrounding dynamics. It can also be integrated into an adaptive cruise control system. This paper presents an omnidirectional lane detection and vehicle motion estimation system. The system employs a multi-camera vision and vehicle detection system to detect the surrounding vehicles. Based on the detected vehicles, the system uses inverse perspective projection and motion estimate techniques to estimate the motion of surrounding vehicles in Cartesian space and to build a bird’s eye view map. To make the vehicle motion estimation more accurate and reliable, Kalman filter, Extended Kalman filter, and particle filter have been used to estimate the motion of surrounding the vehicles. Meanwhile, the system integrates a lane detection technique to detect the lane lines and reconstruct them in Cartesian space to complete the bird’s eye view map with surrounding traffic. Our approach has been successfully validated in real traffic environments by performing experiments with 6 CCD cameras mounted onboard a highway vehicle.