视觉拐弯:转角视野盲区无线感知监测算法

Visual turning:research on the wireless sensing monitoring algorithm for the corner field of the view blind area

驾驶人视野盲区是交通事故频发的一个关键问题。现有的摄像头、转角镜以及传感器在感知环境时容易受到天气或光线不足的影响,从而导致意外事故的发生。因此,基于Wi-Fi网络低功耗、低成本、全天时的泛在感知能力,提出一种基于无线信道状态智能化感知的盲区动态对象监测系统。在类似转角或拐弯的场景下,通过无线感知技术与模式识别,提前监测是否有行人即将进入与行进方向垂直的车辆视野盲区的行为,实现“视觉拐弯”,从而预警来往车辆,避免事故的发生。实验结果表明,所提出的盲区监测系统在5 m范围内的视野盲区中实现行人检测的平均精确度可达到96%,具有鲁棒性与普适性等优势。与传统的监测方式相比,该系统在黑暗环境下能良好地工作,同时不会侵犯人们的隐私,在复杂城市交通网络中具备应用潜力,对社会改善交通状况、保障行人出行安全具有重要意义。

The blind area of a driver′s vision is a key problem in the frequent occurrence of traffic accidents.The existing cameras,corner mirrors and sensors are vulnerable to weather or insufficient light when sensing the environment,resulting in accidents.Therefore,based on the low power consumption,low cost and all-day ubiquitous perception capability of the Wi-Fi network,this paper proposes a blind area dynamic object monitoring system based on intelligent perception of wireless channel status.In scenes similar to corners or turns,wireless sensing technology and pattern recognition are used to monitor whether pedestrians are about to enter the blind area of vehicle vision perpendicular to the travel direction in advance,so as to realize"visual turning",warn passing vehicles and avoid accidents.Experimental results show that the proposed blind area monitoring system can achieve an average accuracy of 96%in the blind area within 5 meters,and has the advantages of robustness and universality.Compared with traditional monitoring methods,the system can work well in the dark environment without violating people′s privacy so that it has application potential in the complex urban traffic network and is of great significance to improving traffic conditions and ensuring pedestrian travel safety.