Computer Vision for Lane Detection and Deviation Warning

• Main Authors: CHEN,BO-ZEN, 陳伯仁
• Other Authors: HAO,MIIN-JONG
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/69ue83

碩士 === 國立高雄科技大學 === 電腦與通訊工程系 === 107 === A vision-based Lane detection technology is important for developing applications in advanced driver-assistance systems (ADAS). In practice, the video images are often interfered by noise and false alarms often occur. In this thesis, an efficient and real-time method for lane detection and deviation warning is proposed. This method is based on the edge detection for image recognition and utilizes the Hough transform to recognize the lane boundaries. In order to improve the lane recognition rate, the video is cut in several pieces and a comparison is performed to further eliminate error information. Moreover, by simply computing the distance between the two different coordinates passing through the lane line, a lane departure warning to the driver is set out for the safety of driving once the measure value exceeds the threshold.