Algorithm Of Gesture Recognition Based On Millimeter Wave Radar

In the fields of computer science and signal processing,gesture recognition has always been a research hotspot.Currently,more mature gesture recognition technologies are based on video images.However,due to the technical characteristics of video images,the existing gesture recognition technologies have some limitations and disadvantages.First,the recognition rate of traditional video image-based gesture recognition technology is seriously affected by light;second,there is also the possibility of privacy leakage.Therefore,designing a new gesture recognition system has important research significance.This paper proposes a gesture recognition system based on millimeter wave radar.Compared with video images,this system combines the advantages of radar equipment,which can solve the lighting effects and privacy leakage problems involved in traditional video image technology.And compared with infrared,laser and other radars,the millimeter-wave radar used in this system has huge advantages in measuring gesture information.Based on this design idea,this subject has studied the algorithm of gesture recognition based on millimeter-wave radar.The millimeter-wave radar is used to collect the echo signals of gestures.The echo signals of the six different gestures of the hand are preprocessed,feature extracted,and finally gesture recognition is completed.The specific work is as follows:The overall framework design of the algorithm system includes millimeter-wave radar sampling data pre-processing module,feature extraction module and gesture recognition classification module.The specific functions of each module are: using millimeter-wave radar equipment to collect information on target gestures,pre-process the sampled data,and remove redundant data information to establish a gesture model;and perform gesture characteristics through optimal time-frequency analysis Extraction;use the classification algorithm to classify different gestures.Implementation of specific algorithms for gesture model establishment and feature extraction.Gesture model A data model based on raw data,which contains rich related gesture information.The most intuitive changes in gestures are the distance,relative speed,and angle between the gesture and the radar.Therefore,the pre-processing method is used to extract these three attributes from the original data as the basis of the gesture model.For feature extraction,a fractional Fourier transform is used to map the gesture model from the time domain to the score domain for feature extraction.Four feature quantities are defined in the score domain.Improvement and Implementation of Gesture Recognition Classification Algorithm.The support vector machine is used to verify that the classification and recognition algorithm can classify and recognize the millimeter-wave radar signals containing gesture information,and provide a reference for the feature selection of subsequent experiments.It is proposed to combine the AdaBoost.M1 algorithm suitable for multi-classification with support vector machine to improve the classification accuracy.Improve the original sample weight update method to reduce the system training sample time.The overall test of the system experiment proved that the system algorithm is feasible and effective.