Human body pose estimation method based on radio frequency heatmap data enhancement

Patent metadata

Field	Value
Publication number	US-12374049-B1
Application number	US-202519192434-A
Country	US
Kind code	B1
Filing date	Apr 29, 2025
Priority date	Dec 3, 2024
Publication date	Jul 29, 2025
Grant date	Jul 29, 2025

Abstract

A human body pose estimation method based on radio frequency heatmap data enhancement is provided, including the following steps: firstly, obtaining the mesh data of human body pose, simulating a radar by using a physical optics method, and obtaining human body mesh features including a radar cross section by irradiating a human body mesh model; secondly, processing the human body mesh features including radar cross section to obtain preliminary simulated radar heatmaps; then inputting the preliminary simulated radar heatmaps into a heatmap conversion network map2map based on U-net, outputting synthetic radar heatmaps, and performing training; finally, combining the synthetic radar heatmaps with real radar heatmaps to construct a mixed data set, obtaining a human body pose skeleton through a human body pose estimation network based on the radar heatmaps, and performing training to complete the human body pose estimation.

First claim

What is claimed is:

1. A human body pose estimation method based on radio frequency heatmap data enhancement, comprising following steps:
step S 1 , obtaining mesh data of human body pose by using a sensor;
step S 2 , simulating a radar by using a physical optics method, constructing a human body mesh model based on the mesh data, and irradiating the human body mesh model to obtain human body mesh features comprising a radar cross section; a specific implementation process is as follows:
S 21 , simulating a plane wave E i emitted by the radar, and irradiating a target human body mesh model; supposing that the target human body mesh model S comprises N triangular facets, and according to the physical optics method and a Stratton-Chu formula, a surface scattering field E n s of one of the irradiated facets S n is defined as follows

E
n
s

=

jk

4
⁢
π

⁢

e

-

jkr
n

r
n

⁢

∫

s
n

(

ZJ

s
n

-

s
×

M

s
n

)

⁢

e

jkr
n

·

(

i
-
s

)

⁢

ds
n

(
1
)

wherein i and S are unit vectors of an incidence direction and an observation direction, respectively; r n is a position vector of a facet S n , and r n is a range from the radar to a target facet S n ; k and Z are wave number and wave impedance in free space, respectively; J s n and M s n respectively represent surface current and surface magnetic current of the facet S n ; e is a natural constant and j is an imaginary unit;
S 22 , then a value σ n of the radar cross section (RCS) of the facet S n at an observation point:

σ
n

=

4
⁢
π
⁢

R
n
2

⁢

❘
"\[LeftBracketingBar]"

E
n
s

❘
"\[RightBracketingBar]"

2

❘
"\[LeftBracketingBar]"

E
n
i

❘
"\[RightBracketingBar]"

2

(
2
)

wherein E n i is an incident electric field of the facet S n ; R n is a range from the facet S n to the observation point;
repeating an above calculation method, and calculating N triangular facets in the human body mesh model S to obtain human body mesh features comprising the RCS;
step S 3 , processing the human body mesh features comprising the radar cross section to obtain preliminary simulated radar heatmaps;
step S 4 , inputting the preliminary simulated radar heatmaps into a heatmap conversion network map2map based on U-net, outputting synthetic radar heatmaps, and performing training; and
step S 5 , combining the synthetic radar heatmaps with real radar heatmaps to construct a mixed data set, obtaining a human body pose skeleton through a human body pose estimation network map2pose based on the radar heatmaps, and performing training to complete the human body pose estimation.

2. The human body pose estimation method based on the radio frequency heatmap data enhancement according to claim 1 , wherein a specific implementation process of the step S 3 is as follows:
S 31 , each of facets has a calculated radar cross section σ n , and coordinates of the facets in a

Assignees

• Univ Hangzhou Dianzi

Inventors

• Wu Yingxiao
• Jiang Zhongmin
• Han Jianping
• Zhou Zhiyuan
• Wang Wenxiang

Classifications

• G01S13/584

  adapted for simultaneous range and velocity measurements · CPC title

• G01S13/89

  for mapping or imaging · CPC title

• G01S13/42

  Simultaneous measurement of distance and other co-ordinates (indirect measurement G01S13/46) · CPC title

• G01S7/412

  based on a comparison between measured values and known or stored values · CPC title

• G01S7/417

  involving the use of neural networks · CPC title