Blind-spot Vehicle Detection with Dynamic and StaticVision Methods

• Main Authors: Shao-zong Ma, 馬紹宗
• Other Authors: Din-chang Tseng
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/88352512681255607686

碩士 === 國立中央大學 === 資訊工程研究所 === 98 === Developing a real-time automotive driver assistant system for safety has emerged wide attention in recent years. When driving on the road, the fields of view beside the host vehicle for drivers are limited. If the driver changes lane without being aware of the objects in the blind-spot area, the potential collision accident may occur. For ensuring the safety of changing lane, our method uses a camera mounted in side-view mirror to capture the image in blind-spot area and detects the vehicle with computer-vision technology. The proposed method offers the blind-spot detection includes defining the detection and decision zone, estimating optical flow, filtering and grouping these estimated optical flow and using the process of tracking and stabilization to accomplish the detection. Considering the situation that optical flow disappears in consecutive tracking process, the proposed method detects the vehicle shadow to keep detecting and tracking. The proposed method also uses the shadow to enhance the detection result generated by optical flow. We apply the proposed detection method to many different situations. In experiments, the detection rate in urban area in daylight is about 95%. The detection rate in suburban area is about 97%. The detection rate in night is about 90%. The detection method operates in Intel? Core 2 Duo? E8400 3.0 GHz CPU, 2GB DDR RAM, Microsoft? Windows 7 has at least 30 frames per seconds.