Human body activity classification method based on weighted group sparse Bayesian learning

Abstract

The invention relates to a human body activity classification method based on weighted group sparse Bayesian learning, and belongs to the technical field of radar and mode recognition. The method comprises the following steps: preprocessing a received human body activity radar echo signal by adopting short-time Fourier transform; performing feature extraction through a principal component analysis method; and using a weighted group sparse Bayesian learning algorithm to carry out sparse coding on a human body activity test sample, and then are classifying and identifying human body activities based on a residual minimum criterion. According to the method, label information of training samples is considered, so that sparse representation coefficients have group structure characteristics, and the classification accuracy is improved; the adopted Bayesian model considers the influence of noise, has good adaptability to the actual environment, and can stably realize human body activity classification; compared with a traditional method, the method has better classification performance under the noisy condition.

Description

technical field [0001] The invention relates to a human body activity classification method based on weighted group sparse Bayesian learning, and belongs to the technical field of radar and pattern recognition. Background technique [0002] With the research and application of the identification and classification of human activities in many fields such as security monitoring and remote health monitoring, radar-based human activity classification has some inherent technologies that do not violate human privacy and can effectively overcome the existence of audio, video, infrared and other life detection. The disadvantages of low cost and easy deployment have become a hot spot for research and development at home and abroad. [0003] Human activities can be regarded as complex targets because various activities will generate complex body movements, and the time-varying trajectories of body parts will be reflected in radar echoes. The radar micro-Doppler feature of human motio...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to deal with the problem that the data collected in the radar human activity classification is affected by noise and cause the classification recognition rate to decline, and proposes a human activity classification method based on weighted group sparse Bayesian learning. The method uses training samples Build a dictionary, use a weighted group sparse Bayesian learning algorithm to sparsely encode test samples, and implement classification based on the minimum residual criterion, which can effectively classify human activities

Images