Radar-Based Non-Contact Audio Recovery Method for Multi-Target Sound Equipment Monitoring
Fengshuo Tian, Zhongxing Wang, Haoyuan Lu, Siyuan Chen, Liuyang Zhang · 2023
Frequency-Modulated Continuous Wave (FMCW) radar demonstrates unconventional capabilities in non-contact monitoring with significant application potential in fault diagnosis, smart homes, and audio recovery. However, current research predominantly focuses on single-target audio recovery, yet leaving the domain of realistic multi-target audio recovery relatively unexplored. This paper introduces a radar-based non-contact audio recovery method for restoring audio temporal signals. The Multiple Signal Classification (MUSIC) algorithm as a crucial component of the adaptive beamforming algorithm is adopted to enhance angular resolution. By enhancing the Minimum Variance Distortionless Response (MVDR)-based adaptive beamforming technique, the reflections from surrounding targets are effectively suppressed, therefore enabling precise monitoring of audio devices. Leveraging the fusion of camera and millimeter-wave radar, along with coordinate transformation and point cloud processing techniques, the audio equipment and background clutter are differentiated. Signal recovery is achieved through bandpass filtering and spectral subtraction. A radar module is affixed to a circular plate and rotated using a servo motor. Radar signal processing is employed to detect object positions, allowing servo motor adjustment of radar measurement range. The proposed method demonstrates robust detection of multi-target audio equipment, contributing to the advancement of non-contact monitoring of multiple target simultaneously.