System and method for detecting and tracking multiple moving targets based on wide-area motion imagery

Patent metadata

Field	Value
Publication number	US-10249047-B2
Application number	US-201815914868-A
Country	US
Kind code	B2
Filing date	Mar 7, 2018
Priority date	Sep 13, 2016
Publication date	Apr 2, 2019
Grant date	Apr 2, 2019

Abstract

Method and system for detecting and tracking multiple moving targets based on wide-area motion imagery are provided. The method includes receiving a task request from a client; sending the task request to a local agent; acquiring to-be-processed images from the client, in response to the task request; calling an Hadoop API and checking a running status of a Hadoop system for an available resource in the Hadoop system; when a resource in the Hadoop system is available, pushing the to-be processed images to a back-end Hadoop distributed file system (HDFS); running tracking algorithms of multiple moving objects associated with the to-be-processed images to provide tracking results of the multiple moving objects; when the Hadoop system is completed in running the tracking algorithms, generating and sending a message to the local agent; and sending the tracking results of the multiple moving objects from the HDFS to the front-end database.

First claim

What is claimed is:

1. A method for detecting and tracking multiple moving targets based on wide-area motion imagery, the method comprising:
receiving, by a message queuing (MQ) platform, a task request from a client in a front-end web layout;
sending, by the MQ platform, the task request to a local agent;
acquiring, by the local agent, to-be-processed images from the client, in response to the task request;
calling, by the local agent, an Hadoop application programming interface (API) and checking a running status of a Hadoop system for an available resource in the Hadoop system;
when a resource in the Hadoop system is available, pushing, by the local agent, the to-be processed images to a back-end Hadoop distributed file system (HDFS);
running, by the Hadoop system, tracking algorithms of multiple moving objects associated with the to-be-processed images to provide tracking results of the multiple moving objects;
when the Hadoop system is completed in running the tracking algorithms, generating and sending a message to the local agent, wherein the local agent further sends the message to the MQ platform; and
sending, by the local agent, the tracking results of the multiple moving objects from the HDFS to the front-end database.

2. The method according to claim 1 , further comprising:
acknowledging, by the MQ platform, the front-end web layout, when the MQ platform receives the message with a completed status from the local agent.

3. The method according to claim 1 , further comprising:
visualizing the tracking results of the multiple moving objects to display on a front-end user interface; and
informing the client for user interaction.

4. The method according to claim 1 , wherein checking the running status of the Hadoop system for the available resource in the Hadoop system includes:
pending the running of the tracking algorithms of the multiple moving objects associated with the to-be-processed images, when the Hadoop system is running another job, and
automatically running the tracking algorithms of the multiple moving objects associated with the to-be-processed images, when the Hadoop system is idle.

5. The method according to claim 1 , wherein:
the tracking results are extracted from image results without keeping the image results,
the tracking results include trajectories of the multiple moving objects, and
each trajectory includes locations of a single object in consecutive to-be-processed images.

6. The method according to claim 5 , wherein:
the image results include a background image, a foreground image, a registration image, and a detection image, and
the image results are stored or retrieved via HDFS.

7. The method according to claim 1 , wherein:
the message sent from the local agent to the MQ platform comprises an assigned task job ID, a task status, attempt times, and a queue repository name.

8. The method according to claim 1 , wherein:
the task request comprises one or more of uploading the to-be-process images and requesting tracking results, and
the task request includes an assigned task job ID.

9. The method according to claim 1 , wherein:
the to-be-process images include images selected from wide-area motion imagery (WAMI) image dataset related to the multiple moving objects.

10. The method according to claim 1 , wherein running, by the Hadoop system, the tracking algorithms of the multiple moving objects includes:
performing a MapReduce implementation for a registration, a background generation, a foreground estimation, and a data association.

11. The method according to claim 10 , wherein:
the data association combines at least positions of a detected multiple moving object in consecutive WAMI frames into target trajectories.

12. The method according to claim 10 , wherein:
MapReduce implementation uses a reducer to perform a summary operation to generate target track IDs and save detection and trajectories information of the multiple moving objects in the HDFS.

13. A system for detecting and tracking multiple moving targets based on wide-area motion imagery, the system comprising:
a message queuing (MQ) platform; a local agent; and a Hadoop system,
wherein:
the message queuing (MQ) platform receives a task request from a client in a front-end web layout, and sends the task request to the local agent,
the local agent acquires to-be-processed images from the client, in response to the task request, calls an Hadoop application programming interface (API) to check a running status of a Hadoop system for an available resource in the Hadoop system, and pushes the to-be processed images to a back-end Hadoop distributed file system (HDFS), when a resource in the Hadoop system is available,
the Hadoop system runs tracking algorithms of multiple moving objects associated with the to-be-processed images to provide tracking results of the multiple moving objects, generates and sends a message to the local agent, when the Hadoop system is completed in running the tracking algorithms, and
the local agent further sends the message to the MQ platform, and sends the tracking results of the multiple moving objects from the HDFS to the front-end database.

14. The system according to claim 13 , wherein:
the MQ platform acknowledges the front-end web layout, when the MQ platform receives the message with a completed status from the local agent.

15. The system according to claim 13 , wherein:
the tracking results of the multiple moving objects is visualized to display on a front-end user interface.

16. The system according to claim 13 , wherein:
the tracking results are extracted from image results without keeping the image results,
the tracking results include trajectories of the multiple moving objects, and
each trajectory includes locations of a single object in consecutive to-be-processed images.

17. The system according to claim 16 , wherein:
the image results include a background image, a foreground image, a registration image, and a detection image, and
the image results are stored or retrieved via HDFS.

18. The system according to claim 13 , wherein:
the message sent from the local agent to the MQ platform comprises an assigned task job ID, a task status, attempt times, and a queue repository name.

19. The system according to claim 13 , wherein:
the task request comprises one or more of uploading the to-be-process images and requesting tracking results, and
the task request includes an assigned task job ID.

20. The system according to claim 13 , wherein:
the to-be-process images include images selected from wide-area motion imagery (WAMI) image dataset related to the multiple moving objects.

Assignees

• Intelligent Fusion Tech Inc

Inventors

• Jia Bin
• Wei Sixiao
• Chen Zhijiang
• Chen Genshe
• Sheaff Carolyn
• Pham Khanh
• Blasch Erik

Classifications

• G06T2207/30241

  Trajectory · CPC title

• H04L67/12

  specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks · CPC title

• G06F9/54 Primary

  Interprogram communication · CPC title

• G06T7/20 Primary

  Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51) · CPC title

• G06F9/547

  Remote procedure calls [RPC]; Web services · CPC title