光学学报, 2019, 39 (5): 0515001, 网络出版: 2019-05-10
基于皮肤表面“振动信号”的多摄像头人体识别定位 下载: 838次
Human Body Recognition and Positioning with Multiple Cameras Based on “Vibration Signals” from Skin Surfaces
多摄像头 人体识别 人体定位 智能视频监控 非接触脉搏 multiple cameras people recognition people positioning intelligent video surveillance non-contact pulse
摘要
提出了一种适应场景广泛的基于皮肤表面“振动信号”的多摄像头人体识别定位方法。提出了同一人体皮肤表面“振动信号”在多摄像头下具有同步性而不同人体皮肤表面 “振动信号”具有差异性的原理,并通过实验进行了验证。基于肤色区域跟踪技术,实现了多摄像头下的人体皮肤表面“振动信号”的获取;基于其差异性和同步性原理及余弦相似度,实现了人体识别;并通过多摄像头坐标计算实现了多摄像头人体定位,最大定位距离误差为61.01 cm。
Abstract
In this study, we propose a method to recognize and position multiple people with multiple cameras based on the “vibration signals” obtained from their skin surfaces to adapt to an extensive range of scenes. Herein, the principles that the “vibration signals” obtained from the same person can be synchronized using multiple cameras and that the “vibration signals” obtained from different people are different are proposed and verified via experiments. The “vibration signals” are obtained from the human skin surfaces using multiple cameras based on skin-color detection and regional tracking. Further, individuals are recognized based on the difference and synchronization principle and the cosine similarity. Several people are positioned with multiple cameras via coordinate calculations, and the maximum positioning distance error is 61.01 cm.
孟濬, 赵夕朦. 基于皮肤表面“振动信号”的多摄像头人体识别定位[J]. 光学学报, 2019, 39(5): 0515001. Jun Meng, Ximeng Zhao. Human Body Recognition and Positioning with Multiple Cameras Based on “Vibration Signals” from Skin Surfaces[J]. Acta Optica Sinica, 2019, 39(5): 0515001.