Method and system for recognition of abnormal behavior

1. A method for recognizing abnormal behavior, the method comprising:

capturing at least one video stream of data on one or more subjects;

extracting body skeleton data from the at least one video stream of data;

classifying the extracted body skeleton data as normal behavior or abnormal behavior;

generating an unknown behavior signal, if the abnormal behavior is a new type of abnormal behavior;

providing the new type of abnormal behavior to an offline analysis module and manually enrolling the new type of abnormal behavior into an incremental hierarchy template database; and

generating an alert, if the abnormal behavior is a known type of abnormal behavior.

2. The method of claim 1 , comprising:

revising the at least one video stream of data to delete video frames, which do not contain human body skeleton information; and

extracting the body skeleton data from the revised sensor video stream data.

3. The method of claim 1 , wherein the step of classifying the extracted body skeleton data comprises:

generating the alert, if the extracted skeleton data does not match a template of normal behavior; and

classifying the extracted skeleton data as the abnormal behavior.

4. The method of claim 1 , comprising:

determining the abnormal behavior using the incremental hierarchy template database, wherein the abnormal behavior is human skeleton data that does not match a normal behavior template within the incremental hierarchy template database.

5. The method of claim 4 , wherein the normal behavior template is a spatial arrangement of skeleton data having an arrangement, which has been previously determined to represent normal behavior.

6. The method of claim 1 , comprising:

recognizing normal or abnormal behavior according to an identified feature, the identified feature comprising at least one frame from the video stream of data; and

classifying the human skeleton data as the normal or abnormal behavior based on the identified feature.

7. The method of claim 6 , wherein the identified feature comprises a multi-frame sequence.

8. The method of claim 1 , comprising:

generating the incremental template hierarchy database based on data clustering using Dendrogram cluster analysis.

9. A system for recognition of abnormal behavior, the system comprising:

an online process module configured to extract body skeleton data from at least one video stream of data;

a behavior recognition module configured to classify a current behavior shown in the at least one video stream of data as abnormal behavior or normal behavior based on the extracted body skeleton data from the at least one video stream of data, the behavior recognition module configured to generate an unknown behavior signal, if the abnormal behavior is a new type of abnormal behavior; and

an offline analysis module configured to provide a user interface for manually enrolling the new type of abnormal behavior and managing the incremental hierarchy template database of the abnormal and normal behaviors, wherein the abnormal behavior consists of the new type of abnormal behavior and known abnormal behavior.

10. The system of claim 9 , wherein the behavior recognition module is configured to determine the abnormal behavior using the incremental hierarchy template database, wherein the abnormal behavior is human skeleton data that does not match a normal behavior template within the incremental hierarchy template database.

11. The system of claim 9 , wherein the incremental hierarchy template database configured to build an abnormal behavior classifier by retraining or clustering an updated incremental hierarchy template database with the abnormal and normal behaviors detected in the sensor video stream of data.

12. The system of claim 9 , wherein the behavior recognition module classifies the current behavior based on the extracted body skeleton data using a template matching system, wherein the extracted body skeleton data is matched to abnormal behavior templates or normal behavior templates, and if the extracted body skeleton data does not match either a normal or abnormal behavior template, the extracted body skeleton data is classified as the new type of abnormal behavior, which is manually enrolled into the incremental hierarchy template database.

13. The system of claim 9 , wherein the at least one video stream is captured using one or more sensor and/or 3-dimensional video cameras.

14. A system for recognition of abnormal behavior, the system comprising:

a video camera configured to capture at least one video stream of data on one or more subjects; and

one or more modules having executable instructions for:

extracting body skeleton data from the at least one video stream of data;

classifying the extracted body skeleton data as normal behavior or abnormal behavior;

generating an unknown behavior signal, if the abnormal behavior is a new type of abnormal behavior;

providing the new type of abnormal behavior to an offline analysis module and manually enrolling the new type of abnormal behavior into the incremental hierarchy template database; and

generating an alert, if the abnormal behavior is a known type of abnormal behavior.

15. The system of claim 14 , comprising:

revising the at least one video stream of data to delete video frames, which do not contain human body skeleton information; and

extracting the body skeleton data from the revised sensor video stream data.

16. A non-transitory computer readable medium containing a computer program having computer readable code embodied therein for recognition of abnormal behavior, comprising:

capturing at least one video stream of data on one or more subjects;

extracting body skeleton data from the at least one video stream of data;

classifying the extracted body skeleton data as normal behavior or abnormal behavior;

generating an unknown behavior signal, if the abnormal behavior is a new type of abnormal behavior;

providing the new type of abnormal behavior to an offline analysis module and manually enrolling the new type of abnormal behavior into the incremental hierarchy template database; and

generating an alert, if the abnormal behavior is a known type of abnormal behavior.

17. The computer readable medium of claim 16 , comprising:

revising the at least one video stream of data to delete video frames, which do not contain human body skeleton information; and

extracting the body skeleton data from the revised sensor video stream data.