Method of detecting position of human body, and system and positioning server using the same

• US 10,175,051 B2
• Filed: 10/11/2016
• Issued: 01/08/2019
• Est. Priority Date: 07/22/2016
• Status: Active Grant

• Current Assignee: Wistron Corporation
• Original Assignee: Wistron Corporation
• Inventors: Chen, Chih-Ming
• Primary Examiner(s): Elchanti, Hussein
• Assistant Examiner(s): Berns, Michael A
• Application Number:
  US15/289,985
• Publication Number:
  US 20180023956A1
• Time in Patent Office: 819 Days
• CPC Class Codes: G01S 17/88 Lidar systems specially adapted for specific applications
• CPC Class Codes: H04W 4/027 using movement velocity, acceleration information
• CPC Class Codes: H04W 4/029 Location-based management or tracking services
• CPC Class Codes: H04W 4/33 for indoor environments, e.g. buildings
• CPC Class Codes: H04W 84/18 Self-organising networks, e.g. ad-hoc networks or sensor networks

Abstract

A human body positioning method, a human body positioning system, and a positioning server are provided, adapted for positioning a human body in a specific space. Best routing paths between a positioning server and a target wireless device among wireless devices are sought according to communication quality of the wireless devices. A device positioning result is estimated according to the best routing paths. A wireless network topology formed by operating wireless devices is established. Voxels of a 3D space image are transmitted into the wireless network topology, and the human body is detected according to the voxels based on serving the operating wireless devices as a neuron of a neural network architecture associated with a body detecting model. When the human body is detected, the human body is positioned at the device positioning result.

Claims

1. A method of detecting a position of a human body in a specific space, wherein a plurality of wireless devices and a positioning server being located in the specific space, the method comprising:

seeking, using a transceiver and a processor executing instructions stored in a memory of the positioning server, a plurality of best routing paths between the positioning server and a target wireless device among the wireless devices according to communication quality of the wireless devices;

identifying, using the processor executing further instructions stored in the memory of the positioning server, a plurality of operating wireless devices among the wireless devices according to the best routing paths, wherein the best routing paths are respectively formed by wireless communication connections between the plurality of operating wireless devices of the wireless devices, the positioning server and the target wireless device, wherein the positioning server and the target wireless device are two ends of each of the best routing paths, and data is transferred between the positioning server and the target wireless device of one best routing path via one or more operating;

wireless devices on that one best routing path;

estimating, using the processor executing further instructions stored in the memory of the positioning server, a device positioning result of the target wireless device according to the best routing paths, wherein the device positioning result indicates a position of the target wireless device in the specific space;

establishing, using the processor executing further instructions stored in the memory of the positioning server, a neural network architecture by setting the operating wireless devices as a plurality of neurons of the neural network architecture for detecting human body image, wherein the neural network architecture comprises a first layer having a plurality of first neurons, and one or more second layers having a plurality of second neurons, and a third layer having third neurons;

capturing, using an image sensor of the target wireless device, a three-dimensional space image of a space around the target wireless device, wherein all original voxels of the three-dimensional space image arc divided into a plurality of partial original voxels by using a processor executing instructions stored in a memory of the target wireless device;

distributing respectively, using a transceiver and the processor executing further instructions stored in the memory of the target wireless device, the plurality of partial original voxels of the three-dimensional space image to the operating transceiver of the first neurons;

respectively performing, using corresponding transceivers and corresponding processors executing corresponding instructions stored in corresponding memories of the plurality of neurons of neural network architecture, image detecting operations to generate a plurality of feature information according to the plurality of partial original voxels and the neural network architecture, wherein a plurality of last generated feature information generated by the third neurons are transmitted to the position server;

determining using the processor executing further instructions stored in the memory of the position server, whether an image of the human body exists in the three-dimensional space image according to the received last generated feature information and a plurality of target feature samples stored in the memory of the position server; and

in response to determining that the image of the human body exists in the three-dimensional space image, determining, using the processor executing further instructions stored in the memory of the position server, that the position of the human body is the device positioning result.

2. The method as claimed in claim 1, wherein the step of performing image detecting operations respectively, using corresponding transceivers and corresponding processors executing corresponding instructions stored in corresponding memories of the plurality of neurons of neural network architecture, to generate a plurality of feature information according to the plurality of partial original voxels and the neural network architecture comprises:

respectively performing, using corresponding processors executing corresponding instructions stored in corresponding memories of the first neurons, first image detecting operations to the plurality of partial original voxels to generate a plurality of first feature information, and transmitting respectively, using corresponding transceivers of the first neurons, the plurality of first feature information to the second neurons;

respectively performing, using corresponding processors executing corresponding instructions stored in corresponding memories of the second neurons, second image detecting operations to the first image information to generate a plurality of second feature information, and transmitting respectively, using corresponding transceivers of the second neurons, the plurality of second feature information to the third neurons; and

respectively performing, using corresponding processors executing corresponding instructions stored in corresponding memories of the third neurons, third image detecting operations to the second image information to generate a plurality of third feature information, and transmitting respectively, using corresponding transceivers of the third neurons, the plurality of third feature information to the position server, wherein the third feature information is the last generated feature information;

determining, using the processor executing further instructions stored in the memory of the position server, that the image of the human body exists in the three-dimensional space image when one of the third feature information matches one of the plurality of target feature samples.

3. The method as claimed in claim 2, wherein the neural network architecture is a deep neural network, and the first layer, the second layer and the third layer belong to hidden layers of the deep neural network.

4. The method as claimed in claim 1, wherein the neural network architecture is a recurrent neural network, and each of the operating wireless devices is set as a dynamic neuron of the recurrent neural network, and the operating wireless devices detect the image of the human body by using the three-dimensional space image of the space corresponding to a time point and another three-dimensional space image of the space corresponding to another time point.

5. The positioning method as claimed in claim 1, wherein the step of seeking the best routing paths between the positioning server and the target wireless device among the wireless devices according to communication quality of the wireless devices comprises:

establishing, using the corresponding transceivers and the corresponding processors executing corresponding further instructions stored in the corresponding memories of the wireless devices, the wireless communication connections between the wireless devices according to wireless connection capabilities of the wireless devices, such that the wireless devices form an ad-hoc network; and

obtaining, using the transceiver and the processor executing further instructions stored in the memory of the positioning server, communication quality parameters of the wireless communication connections, and seeking the best routing paths between the positioning server and the target wireless device among the wireless devices according to the communication quality parameters of the wireless communication connections.

6. The method as claimed in claim 1, wherein the step of estimating the device positioning result of the target wireless device according to the best routing paths comprises:

obtaining, using the processor executing further instructions stored in the memory of the positioning server, positioning locations respectively corresponding to the best routing paths; and

performing, using the processor executing further instructions stored in the memory of the positioning server, a statistical operation to the positioning locations to obtain the device positioning result.

7. The method as claimed in claim 1, further comprising:

obtaining, using the processor executing further instructions stored in the memory of the positioning server, semantic features from the last generated feature information; and

comparing, using the processor executing further instructions stored in the memory of the positioning server, the semantic features with semantic samples of the target feature samples to recognize that the image of the human body is one of the target feature samples.

8. The method as claimed in claim 7, wherein the step of obtaining the semantic features of the human body according to the another neural network architecture for detecting human body image comprises:

obtaining, using the processor executing further instructions stored in the memory of the target wireless device, gait vectors of the human body, and transmitting the gait vectors into the another neural network architecture to obtain the semantic features according to the gait vectors,

wherein each of the operating wireless devices is set as a neuron of the another neural network architecture.

9. A system of detecting a position of a human body in a specific space, the system comprises:

a positioning server, located in the specific space, wherein the positioning server comprises a server transceiver, a server processor and a server memory; and

a plurality of wireless devices, located in the specific space,

wherein the server processor executes instructions stored in the server memory to seek a plurality of best routing paths between the positioning server and a target wireless device among the wireless devices according to communication quality of the wireless devices, wherein the target wireless device comprises an image sensor a target transceiver a target processor and a target memory,

wherein the server processor executes further instructions stored in the server memory to identify a plurality of operating wireless devices among the wireless devices according to the best routing paths wherein each of the operating wireless devices comprise an operating transceiver, an operating processor and an operating memory, wherein the best routing paths are respectively formed by wireless communication connections between the plurality of operating wireless devices of the wireless devices, the positioning server and the target wireless device, wherein the positioning server and the target wireless device are two ends of each of the best routing paths, and data is transferred between the positioning server and the target wireless device of one best routing path via one or more operating wireless devices on that one best routing path,

wherein the server processor executes further instructions stored in the server memory to estimate a device positioning result of the target wireless device according to the best routing paths, wherein the device positioning result indicates a position of the target wireless device in the specific space,

wherein the server processor executes further instructions stored in the server memory to establish a neural network architecture by setting the operating-wireless devices as a plurality of neurons of the neural network architecture for detecting human body image, wherein the neural network architecture comprises a first layer having a plurality of first neurons, and one or more second layers having a plurality of second neurons, and a third layer having third neurons,

wherein the image sensor is configured to capture a three-dimensional space image of a space around the target wireless device wherein all the original voxels of the three-dimensional space image are divided into a plurality of partial original voxels by the target processor executing instructions stored in the target memory,

wherein the target processor executes further instructions stored in the target memory to distribute respectively the plurality of partial original voxels of the three-dimensional space image to operating transceivers of the first neurons via the target transceiver,

wherein corresponding operating processors execute corresponding instructions stored in corresponding operating memories of the plurality of neurons of neural network architecture to respectively perform image detecting operations to generate a plurality feature information according to the plurality of partial original voxels and the neural network architecture, wherein a plurality of last generated feature information generated by the third neurons are transmitted to the position server,

wherein the server processor executing further instructions stored in the server memory to determine whether an image of the human body exists in the three-dimensional space image according to the received last generated feature information and a plurality of target feature samples stored in the server memory,

wherein in response to determining that the image of the human body exists in the three-dimensional space image, the server processor executing further instructions stored in the server memory to determine that the position of the human body is the device positioning result.

10. The system as claimed in claim 9,

wherein corresponding operating processors of the first neurons are configured to respectively perform first image detecting operations to the plurality of partial original voxels to generate a plurality of first feature information, and transmit the plurality of first feature information to the second neurons via corresponding operating transceivers of the first neurons,

wherein corresponding operating processors of the second neurons are configured to perform second image detecting operations to the first image information to generate a plurality of second feature information, and transmit the plurality of second feature information to the third neurons via corresponding operating transceivers of the second neurons,

wherein corresponding operating processors of the third neurons are configured to perform third image detecting operations to the second image information to generate a plurality of third feature information, and transmit respectively the plurality of third feature information to the position server via corresponding transceivers of the third neurons, wherein the third feature information is the last generated feature information,

wherein the server processor is configured to determine that the image of the human body exists in the three-dimensional space image when one of the third feature information matches one of the plurality of target feature samples.

11. The system as claimed in claim 10, wherein the neural network architecture is a deep neural network, and the first layer devices, the second layer devices and the third layer devices belong to hidden layers of the deep neural network.

12. The system as claimed in claim 9, wherein the neural network architecture is a recurrent neural network, and each of the operating wireless devices is set as a dynamic neuron of the recurrent neural network, and the operating wireless devices detect image of the human body by using the three-dimensional space image of the space corresponding to a time point and another three-dimensional space image of the space corresponding to another time point.

13. The system as claimed in claim 9, wherein the corresponding transceivers of the wireless devices establish wireless communication connections between the wireless devices according to wireless connection capabilities of the wireless devices, such that the wireless devices form an ad-hoc network,

wherein the server processor obtains communication quality parameters of the wireless communication connections via the server transceiver, and seeks the best routing paths between the positioning server and the target wireless device among the wireless devices according to the communication quality parameters of the wireless communication connections.

14. The system as claimed in claim 9, wherein the server processor obtains positioning locations respectively corresponding to the best routing paths, and performs a statistical operation to the positioning locations to obtain the device positioning result.

15. The system as claimed in claim 9, wherein based on the neurons of the another neural network architecture, the server processor obtains semantic features from the last generated feature information,

the server processor compares the semantic features with semantic samples of target feature samples to recognize that the image of the human body is one of the target feature sample.

16. The system as claimed in claim 15, wherein the positioning server establishes a sample database comprising the semantic samples of the target feature samples.

17. A positioning server of detecting a position of a human body in a specific space, wherein a plurality of wireless devices are located in the specific space to form an ad-hoc network, the positioning server comprising:

a server transceiver; and

a server processor, coupled to the server transceiver, and server processor is configured to seek a plurality of best routing paths between the positioning server and a target wireless device among the wireless devices according to communication quality of the wireless devices, wherein the target wireless device comprises an image sensor, a target transceiver, and a target processor,

wherein the server processor is configured to identify a plurality of operating wireless devices among the wireless devices according to the best routing paths,

wherein each of the operating wireless devices comprise an operating transceiver and an operating processor, wherein the best routing paths are respectively formed by wireless communication connections between the plurality of operating wireless devices of the wireless devices, the positioning server and the target wireless device, wherein the positioning server and the target wireless device are two ends of each of the best routing paths, and data is transferred between the positioning server and the target wireless device of one best routing path via one or more operating wireless devices on that one best routing path,

wherein the server processor is configured to estimate a device positioning result of the target wireless device according to the best routing paths, and establishes a wireless network topology formed by the operating wireless devices, wherein the device positioning result indicates a position of the target wireless device in the specific space,

wherein the server processor is configured to establish a neural network architecture by setting the operating wireless devices as a plurality of neurons of the neural network architecture for detecting human body image, wherein the neural network architecture comprises a first layer having a plurality of first neurons, and one or more second layers having a plurality of second neurons, and a third layer having third neurons,

wherein the image sensor of the target wireless device is configured to capture a three-dimensional space image of a space around the target wireless device, wherein all original voxels of the three-dimensional space image are divided into a plurality of partial original voxels by the target processor executing instructions stored in the target memory

wherein the target processor is configured to distribute respectively the plurality of partial original voxels of the three-dimensional space image to the operating transceiver of the first neurons via the target transceiver,

wherein corresponding operating processors of the neurons of neural network architecture are configured to respectively perform image detecting operations to generate a plurality of feature information according to the plurality of partial original voxels and the neural network architecture, wherein a plurality of last generated feature information generated by the third neurons are transmitted to the position server,

wherein the server processor is configured to determine whether an image of the human body exists in the three-dimensional space image according to received last generated feature information and a plurality of target feature samples stored in the server memory,

wherein in response to determining that the image of the human body exists in the three-dimensional space image, the server processor is configured to determine that the position of the human body is the device positioning result.

18. The positioning server as claimed in claim 17, wherein wireless communication connections between the wireless devices are established according to wireless connection capabilities of the wireless devices, and the server processor obtains communication quality parameters of the wireless communication connections via the server transceiver, and seeks the best routing paths between the positioning server and the target wireless device among the wireless devices according to the communication quality parameters of the wireless communication connections.

19. The positioning server as claimed in claim 17, wherein the server processor obtains positioning locations respectively corresponding to the best routing paths, and performs a statistical operation to the positioning locations to obtain the device positioning result.

20. The positioning server as claimed in claim 17, wherein the server processor establishes a sample database comprising semantic samples of target feature samples, and compares semantic features of the last generated feature information with the semantic samples of the target feature samples to recognize that the image of the human body is one of the human body samples.