摘要
针对现有室内定位技术存在的问题,文中提出一种精度高、成本低且易用性高的室内定位系统。系统采集并处理监控摄像头中的视频信号,用计算机视觉算法从视频流中提取并追踪行人的位置信息,通过基于卷积神经网络的物体追踪算法处理遮挡状况。通过智能手机内嵌传感器采集的运动特征并与心中的行人特征对比,系统可以在视频画面内的诸多行人中准确识别出持有智能手机的定位服务发起者,并报告实时的定位信息,定位平均误差可达到三十厘米以下。
In order to solve the problems of indoor positioning techniques,in this paper we propose a novel indoor positioning system with high accuracy,low overhead and ease of use.The proposed system collect and process the video stream from surveillance camera,then extract pedestrians out of the video image and track their position.We apply a convolutional neural network based object tracking algorithm to handle occlusion.The system extract gait features from both video and smartphone,and recognize target pedestrian in the video space.The average positioning error is smaller than 30 cm.
作者
张久鑫
ZHANG Jiu-xin(Shanghai Institute of Microsystem and Information Technology,Chinese Academy of Sciences,Shanghai 200050,China;School of Information Science and Technology,ShanghaiTech University,Shanghai 201210,China;;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《电子设计工程》
2018年第24期140-143,149,共5页
Electronic Design Engineering
