摘要
设计了一种基于深度学习行人检测算法的交通安全智能警示系统。该系统通过计算图像的灰度和模糊度来进行红外切换,再通过相应的YOLOv3算法检测行人目标,并对行人目标候选框提取局部二值模式特征(LBP)和多层感知器(MLP)分类。处理器在检测到行人后控制光电模块示警,视觉上警示行驶车辆减速慢行;同时控制扬声器播报语音,听觉上提醒行人小心通过;处理器会控制FM无线发射器,使在一定距离内且相同频道的车载收音机播放"前方有行人通过"的提示音,提醒过路司机。经过实际测试,此系统有效起到了对行人和过路司机的提示作用,具有应用价值。
In this paper, we proposed an intelligent warning system based on pedestrian detection, which could be used especially in transportation industry. The system performed infrared switching by calculating grayscale and blur of the images;And then detected pedestrians through the corresponding YOLOv3 algorithm;Meanwhile, extracted the local binary pattern feature(LBP)and classified from the candidate boxes by multi-layer perception(MLP). After the processor(MCU)detected some pedestrian,it would control photoelectric modules to warn some driving vehicles to slow down, control loudspeaker to broadcast, so as to remind pedestrians to pass carefully, and control FM radio transmitter so that the car radio using the same channel would play the warning voice of "pedestrians ahead".After practical test, the system has played an effective warning role in prompting pedestrians and drivers, which has application value.
作者
柳黎
何伍斌
杨赛
解铮
LIU Li;HE Wu bin;YANG Sai;XIE Zheng(Jiangsu Suchness Geographical Information Tech,Suqian 223800,China;Jiangsu Suchness Mathematics Institute,Suqian 223800,China)
出处
《电子设计工程》
2020年第4期96-99,104,共5页
Electronic Design Engineering
基金
江苏省宿迁市科技项目(SC201801)。
