Dynamic Gesture Recognition with a Terahertz Radar Based on Range Profile Sequences and Doppler Signatures

Area of Science:

• Radar Systems and Signal Processing
• Human-Computer Interaction
• Electromagnetics

Background:

• Terahertz (THz) radar operates at higher frequencies (0.1-10 THz) than microwaves, offering unique sensing capabilities.
• While THz radar excels in automatic target recognition using high-resolution range profiles (HRRP) and Doppler signatures, its application in dynamic gesture recognition remains underexplored.

Purpose of the Study:

• To propose and validate a novel dynamic gesture recognition system leveraging multi-modal signals from terahertz radar.
• To explore the potential of THz radar for advanced human-computer interaction through gesture analysis.

Main Methods:

• Acquisition of multi-modal signals, specifically HRRP sequences and Doppler signatures, from terahertz radar echoes.
• Development of a feature extraction model based on the location parameter estimation of scattering centers, considering electromagnetic scattering characteristics.
• Application of Dynamic Time Warping (DTW), extended for multi-modal signals, for gesture classification.

Main Results:

• Successful collection and analysis of ten distinct gesture types using the terahertz radar system.
• Experimental validation demonstrating a gesture recognition rate exceeding 91% for the proposed system.
• Confirmation of the system's efficacy in accurately identifying dynamic gestures.

Conclusions:

• The developed terahertz radar system demonstrates significant potential for accurate dynamic gesture recognition.
• Multi-modal signal analysis, combined with scattering center feature extraction and DTW, is effective for gesture classification in the terahertz regime.
• This research opens avenues for advanced applications of terahertz radar in human-computer interaction and beyond.