Three-dimensional feature extraction from frequency modulated continuous wave radar signals
View Patent ↗Motion-related 3D feature extraction by receiving a set of sequential radar signal data frames associated with a subject, determining radar signal amplitude and phase for each data frame, determining radar signal phase changes between sequential data frames, and extracting, by a trained machine learning model, one or more three-dimensional features from the sequential radar signal data frames according to the radar signal amplitude and the radar signal phase changes between sequential data frames.
1. A computer implemented method for three-dimensional feature extraction from sequential radar signal data frames, the method comprising:
receiving a set of sequential radar signal data frames;
determining radar signal amplitude and phase for each data frame;
determining radar signal phase changes between sequential data frames;
transforming the radar signal amplitude and phase change from a previous frame for a frame to a four-dimensional tensor including spatial location as well as magnitude and phase velocity data;
and
extracting, by a trained machine learning model, one or more three-dimensional features from the sequential radar signal data frames according to the four-dimensional tensor including spatial location as well as magnitude and phase velocity data.
2. The computer implemented method according to claim 1 , wherein the three-dimensional features comprise temporal and directional three-dimensional subject motion features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
3. The computer implemented method according to claim 1 , wherein the three-dimensional features comprise volumetric features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
4. The computer implemented method according to claim 1 , wherein determining radar signal phase changes between sequential data frames comprises determining radar signal phase velocity between sequential frames.
5. The computer implemented method according to claim 1 , wherein the radar signal data frames comprise a signal bandwidth of about 1 GHz.
6. The computer implemented method according to claim 1 , further comprising providing the three-dimensional features to an event recognition system.
7. The computer implemented method according to claim 1 , wherein the trained machine learning model comprises one or more long short-term memory components.
8. A computer program product for three-dimensional feature extraction from sequential radar signal data frames, the computer program product comprising one or more computer readable storage devices and program instructions collectively stored on the one or more computer readable storage devices, the program instructions comprising:
program instructions to receive a set of sequential radar signal data frames;
program instructions to determine radar signal amplitude and phase for each data frame;
program instructions to determine radar signal phase changes between sequential data frames;
program instructions to transform the radar signal amplitude and phase change from a previous frame for a frame to a four-dimensional tensor including spatial location as well as magnitude and phase velocity data; and
program instructions to extract, by a trained machine learning model, one or more three-dimensional features from the sequential radar signal data frames according to the four-dimensional tensor including spatial location as well as magnitude and phase velocity data.
9. The computer program product according to claim 8 , wherein the three-dimensional features comprise temporal and directional subject motion features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
10. The computer program product according to claim 8 , wherein the three-dimensional features comprise volumetric features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
11. The computer program product according to claim 8 , wherein determining radar signal phase changes between sequential data frames comprises determining radar signal phase velocity between sequential frames.
12. The computer program product according to claim 8 , wherein the radar signal data frames comprise a signal bandwidth of about 1 GHz.
13. The computer program product according to claim 8 , the stored program instructions further comprising program instructions to provide the three-dimensional features to an event monitoring system.
14. The computer program product according to claim 8 , wherein the trained machine learning model comprises one or more long short-term memory components.
15. A computer system for three-dimensional feature extraction from sequential radar signal data frames, the computer system comprising:
one or more computer processors;
one or more computer readable storage devices; and
stored program instructions on the one or more computer readable storage devices for execution by the one or more computer processors, the stored program instructions comprising:
program instructions to receive a set of sequential radar signal data frames;
program instructions to determine radar signal amplitude and phase for each data frame;
program instructions to determine radar signal phase changes between sequential data frames;
program instructions to transform the radar signal amplitude and phase change from a previous frame for a frame to a four-dimensional tensor including spatial location as well as magnitude and phase velocity data; and
program instructions to extract, by a trained machine learning model, one or more three-dimensional features from the sequential radar signal data frames according to the four-dimensional tensor including spatial location as well as magnitude and phase velocity data.
16. The computer system according to claim 15 , wherein the three-dimensional features comprise temporal and directional subject motion features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
17. The computer system according to claim 15 , wherein the three-dimensional features comprise volumetric subject motion features according to the radar signal amplitude and radar signal phase changes between sequential data frames.
18. The computer system according to claim 15 , wherein determining radar signal phase changes between sequential data frames comprises determining radar signal phase velocity between sequential frames.
19. The computer system according to claim 15 , wherein the radar signal data frames comprise a signal bandwidth of about 1 GHz.
20. The computer system according to claim 15 , the stored program instructions further comprising program instructions to provide the three-dimensional features to an event recognition system.