Research And Application Of People Counting System Based On Feature Extraction Of Wireless Signal

People counting is playing an important role in military operations,emergency search and rescue,and public safety.Wearable systems need to be worn all the time,and the application scenarios are very limited.Visible light machine vision is greatly affected by issues such as occlusion,changes in light intensity,and privacy protection.Therefore,it is necessary to detect the number of people living by using radio waves that can penetrate walls during day and night and will not cause privacy leakage.Wi Fi can not only be used to transmit data,but the received Wi Fi signal will carry characteristic information that reflects the state of the living body.Currently,Wi Fi-based people counting technology mainly has bottlenecks such as low resolution,large signal noise,multiple echo reflections and lack of concentration,and strong environmental dependence,making it difficult for people to directly understand what is happening through wireless signals.Therefore,it is urgent to break through the key technologies of wireless signal perception,study the expression mechanism and feature extraction methods of wireless signals for the number of people,build a living-body number detection model,form a new generation of non-contact active number-sensing capabilities,and achieve better scene understanding and update Smart space environment.Aiming at the actual problem of people counting,this paper abstracts it as the problem of image expression and modeling of wireless signals,and integrates signal analysis theory and data-driven methods to carry out research on new sensing technologies.The specific work is as follows:First,build a Wi Fi-CSI signal acquisition system to realize automatic acquisition,further analyze the temporal and spatial distribution characteristics of Wi Fi signals,and propose an adaptive extraction method for effective CSI signals.Second,establish a quantitative mapping model between the effective CSI signal and the radio image.A method of expressing one-dimensional time series CSI signals using two-dimensional intuitive image sequences is proposed to realize the "visible" of "invisible" wireless signals.Third,build a people counting model that integrates multiple models.This paper builds a model from the perspective of image analysis,studies the people counting model based on CSI radio images,and establishes a people counting program based on CSI signal analysis from both time and space directions,and studies the people counting method based on the SE-RF-Res Net fusion model.,The method achieves 97% accuracy in an environment of zero to five people.Fourth,develop a people counting system based on Wi Fi-CSI signals.This article designs the overall architecture of the people counting system,divides the entire people counting system into two parts,the Wi Fi-CSI signal acquisition system and the real-time people counting system,and elaborates the architecture design and development process of the two.