Multi-person Tracking And Gait Recognition Based On Millimeter Wave Sensing

At present,biometric technology has been widely used in people's daily life,such as unlocking and payment.Specific technologies include fingerprint recognition and face recognition.However,these biometric technologies have obvious drawbacks.They can only be used for biometric identification at close range or touch.When a person is at a long distance and cannot touch,these technologies will not be available.Gait recognition can perfectly solve the above drawbacks.Previous,gait recognition research work was mainly based on image recognition.The disadvantage of image recognition can only be used when the light is good and thing cannot be covered.With the continuous development and maturity of 5G technology,millimeter waves will continue to be applied to smart devices.High resolution is the natural advantage of millimeter wave,which makes millimeter wave can be used in the field of intelligent perception.In this article,we propose a method for multi-person tracking and gait recognition based on millimeter wave perception in a scene where multiple people coexist.Through the use of millimeter wave sensing technology to perceive the gait of pedestrians in a scene where multiple people coexist,and obtain 3D point cloud gait data of pedestrians.Because the data is in a scene where multiple people coexist,this paper proposes a method for separating and tracking multiple people's gait data based on clustering and multi-target matching algorithms.Through this method,each single person's gait data is obtained.For each single person's gait data,this paper designs a gait recognition algorithm based on deep learning.Experiments show that the algorithm achieves 90%accuracy in a single person scenario and 88%accuracy in a five people coexistence scenario.Finally,the method proposed in this paper is implemented as a basic millimeter wave sensing multi-person tracking and gait recognition system,which achieves a good gait recognition effect.