研究目的
Investigating the feasibility of recognizing human hand gestures using micro-Doppler signatures measured by Doppler radar with a deep convolutional neural network (DCNN).
研究成果
The study investigated the feasibility of a proposed method that classifies human hand gestures using micro-Doppler signatures with a DCNN. The classification accuracy of the proposed method was found to be 85.6% for ten gestures. With seven gestures, the accuracy increased to 93.1%. For robust and practical operation, data from diverse scenarios should be included in the measurement process.
研究不足
Micro-Doppler signatures can vary depending on aspect angle and distance to the radar. For robust and practical operation, data from diverse scenarios should be included in the measurement process. In addition, multiple human subjects should be measured to construct a user-independent classifier.
1：Experimental Design and Method Selection:
Employed Doppler radar to obtain micro-Doppler signatures of ten hand gestures from a single participant. A DCNN was used to classify the spectrograms.
2：Sample Selection and Data Sources:
Ten hand gestures were measured using Doppler radar and their spectrograms analyzed. Each gesture was measured 50 times, resulting in 500 pieces of data in total.
3：List of Experimental Equipment and Materials:
Bumblebee Doppler radar (Samraksh Co. Ltd.), which operates at 5.8 GHz, was employed.
4：8 GHz, was employed.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: Hand gestures were executed in the main lobe of the radar antenna. Spectrograms of finger motions were observed through short-time fast Fourier transform (FFT).
5：Data Analysis Methods:
A DCNN was employed to classify the spectrograms, with 90% of the data utilized for training and the remaining 10% for validation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
