Sleep posture classification method based on millimeter wave radar multi-view data fusion

Abstract

The embodiment of the present disclosure provides a sleep posture classification method based on millimeter wave radar multi-view data fusion, which belongs to the technical field of data recognition, and specifically comprises: performing distance fast Fourier transform on each frame of original radar data; performing Doppler fast Fourier transform processing on the frequency spectrum information in the distance dimension; selecting and splicing the data after Doppler threshold processing to generate horizontal angle data and elevation angle data; performing angle fast Fourier transform on the horizontal angle data, and performing angle fast Fourier transform processing on the elevation angle data; taking a preset number of frames of data as a recognition window for the original radar data; using a feature extractor to extract features of multi-view radar sleep posture data respectively, integrating the features of two views through a global multi-view fusion mechanism, forming a fusion feature representation and inputting the multi-layer perception machine for individual recognition and sleep posture recognition. Through the scheme of the present disclosure, the recognition efficiency, accuracy and adaptability are improved.

Images