Multi-person pose recognition system using a zigbee wireless sensor network

• US 7,742,894 B2
• Filed: 03/14/2008
• Issued: 06/22/2010
• Est. Priority Date: 12/24/2007
• Status: Active Grant

• Current Assignee: National Chiao Tung University (Government of The Republic of China)
• Original Assignee: National Chiao Tung University (Government of The Republic of China)
• Inventors: Chen, Chun-Wei, Song, Kai-Tai
• Primary Examiner(s): Wachsman; Hal D
• Assistant Examiner(s): HENSON, MISCHITA L
• Application Number:
  US12/048,678
• Publication Number:
  US 20090161915A1
• Time in Patent Office: 830 Days
• Field of Search: 702/40, 702/57, 702/66, 702/82, 702 94- 97, 702/104, 702/142, 702149-150, 702189-191, 73510-511, 340/669, 340/686.1, 340/573.1, 340/573.7, 340/699
• US Class Current: 702/150
• CPC Class Codes: A61B 2562/0219 Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
• CPC Class Codes: A61B 5/1038 Measuring plantar pressure during gait
• CPC Class Codes: A61B 5/1117 Fall detection
• CPC Class Codes: A61B 5/1123 Discriminating type of movement, e.g. walking or running A61B5/1116, A61B5/112 take precedence
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00

Abstract

In the present invention, a multi-person pose recognition system has been developed. This system includes a body pose detection module, a CC2420DBK board and a multi-person pose monitoring software module. The body pose detection module includes a triaxial accelerometer, a Zigbee chip and an 8-bit microcontroller. Several body pose detection modules and the CC2420DBK board form a Zigbee wireless sensor network (WSN). The CC2420DBK board functions as the receiver of the Zigbee WSN and communicates with a robot onboard computer or a host computer through a RS-232 port. The multi-person pose monitoring software monitors and records activities of multiple users simultaneously. The present invention provides a pose recognition algorithm by combining time-domain analysis and wavelet transform analysis. This algorithm has been implemented in the microcontroller of a body pose estimation module. Through the algorithm, the system can recognize seven body poses: falling, standing, sitting, lying, walking, going upstairs and going downstairs.

Claims

1. A method for estimating a body pose in a body pose detection module provided with a triaxial accelerometer within a wireless sensor network of a plurality of body pose detection modules arranged to detect and simultaneously monitor body poses of multiple users, the method comprising steps of:

performing signal pre-processing of a triaxial acceleration signal readout from the triaxial accelerometer at the body pose detection module attached to a user body so as to abstract a dynamic acceleration and a static acceleration of the triaxial acceleration signal;

determining whether the dynamic acceleration of the triaxial acceleration signal is abnormal indicating that a body pose is a falling pose; and

when the dynamic acceleration of the triaxial acceleration signal is not abnormal, analyzing the dynamic acceleration of the triaxial acceleration signal to determine whether a body pose is one of a static pose and a dynamic pose,

wherein the signal preprocessing is performed by;

sampling from the triaxial acceleration signal readout from the triaxial accelerometer, each of sensing axes collecting 256 acceleration data, with a sampling frequency of 128 Hz and performing the signal pre-processing once at every two seconds;

subjecting acceleration data of the sensing axes of the triaxial acceleration signal to an eight-layer Haar wavelet transform;

maintaining a wavelet transform coefficient of the eight-layer Haar wavelet transform with a lowest frequency, while omitting other wavelet transform coefficients;

converting the wavelet transform coefficient of the lowest frequency into a time domain to represent the static acceleration of the triaxial acceleration signal; and

subtracting the static acceleration from the triaxial acceleration signal to obtain the dynamic acceleration of the triaxial acceleration signal.

2. The method as claimed in claim 1, further comprising a step of determining the body pose as a standing, sitting or lying pose in accordance with an inclination angle of a user trunk, if the body pose is determined as a static pose.

3. The method as claimed in claim 2, wherein the inclination angle of the body trunk is determined by the static acceleration sensed along a Y sensing axis of the triaxial accelerometer, and which one of the standing, sitting or lying pose the body pose pertains to is determined by steps of:

determining the body pose as a standing pose if g·

cos(20°

)<

g·

cos(θ

)≦

g·

cos(0°

);

determining the body pose as a sitting pose if g·

cos(60°

)<

g·

cos(θ

)≦

g·

cos(20°

); and

determining the body pose as a lying pose if g·

cos(θ

)≦

g·

cos(60°

);

where g is an acceleration of gravity of the earth, θ

is an included angle between the acceleration of gravity of the earth, and the Y sensing axis of the axial accelerometer, g·

cos(θ

) represents the static acceleration sensed along the Y sensing axis and is a cosine component of the acceleration of gravity of the earth.

4. The method as claimed in claim 1, wherein the static acceleration is a component of the acceleration of gravity of the earth.

5. The method as claimed in claim 1, further comprising a step of detecting if a step movement exists through signal preprocessing for a step movement detection and a step movement determination, if the body pose is determined as a dynamic pose.

6. The method as claimed in claim 5, wherein the signal preprocessing for step movement detection and step movement determination further comprises:

using a local minimum in a Z-axis acceleration signal of the body pose to be determined as an initial characteristic value of the step movement;

fetching two consecutive steps as one cycle and providing cyclized acceleration data up to a preset amount for the calculation of the wavelet transform;

performing a two-layer wavelet transform to the square of the Z-axis acceleration signal; and

locating a position of a local maximum by virtue of wavelet coefficients after the wavelet transform and then estimating an approximate position where the local maximum in the time domain occurs.

7. The method as claimed in claim 5, further comprising steps of:

if the step movement is detected,

applying a wavelet transform to the triaxial acceleration signal;

analyzing wavelet coefficients of an energy distribution band while the user body is walking normally;

calculating characteristic parameters in accordance with the wavelet coefficients; and

determining, based on the step movement, if the body pose is a walking pose, or pose of going upstairs or downstairs;

otherwise

classifying the step movement as an exercising state.

8. The method as claimed in claim 7, wherein the characteristic parameters comprise:

an RTF for determining if the step movement is a body pose while going upstairs, when a RTF value exceeds a preset threshold; and

an RVF for determining if the step movement is a body pose while going downstairs, when a RVF value exceeds a preset threshold.

9. The method as claimed in claim 8, wherein the RTF is expressed by the equation as follows:

$\mathrm{RTF}=\frac{{<br><br>a_z<br><br>}_2^2}{\underset{j=3}{\overset{6}{\sum <br><br>}}<br><br>d_{\mathrm{zj}}}$

where a_z is a Z-axis acceleration, and d_z is a Z-axis wavelet decomposition coefficient.

10. The method as claimed in claim 8, wherein the RVF is expressed by the equation as follows:

$\mathrm{RVF}=\frac{\underset{j=3}{\overset{6}{\sum <br><br>}}<br><br>d_{\mathrm{yj}}}{\underset{j=3}{\overset{6}{\sum <br><br>}}<br><br>d_{\mathrm{zj}}}$

where d_y is a Y-axis wavelet decomposition coefficient, and d_z is a Z-axis wavelet decomposition coefficient.