ABNORMAL BEHAVIOR DETECTION SYSTEM AND METHOD USING AUTOMATIC CLASSIFICATION OF MULTIPLE FEATURES

• US 20120134532A1
• Filed: 11/29/2010
• Published: 05/31/2012
• Est. Priority Date: 06/08/2010
• Status: Active Grant

• Current Assignee: Gorilla Technology UK Limited
• Original Assignee: Gorilla Technology, Inc. (Gorilla Science & Technology, Inc.)
• Inventors: NI, SZE-YAO, LAN, YUANG-TZONG, LIN, TOM C.I., CHEN, YI-WEI
• Granted Patent: US 8,885,929 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G08G 1/0175 by photographing vehicles, e.g. when violating traffic rules

Abstract

Described herein are a system and a method for abnormal behavior detection using automatic classification of multiple features. Features from various sources, including those extracted from camera input through digital image analysis, are used as input to machine learning algorithms. These algorithms group the features and produce models of normal and abnormal behaviors. Outlying behaviors, such as those identified by their lower frequency, are deemed abnormal. Human supervision may optionally be employed to ensure the accuracy of the models. Once created, these models can be used to automatically classify features as normal or abnormal. This invention is suitable for use in the automatic detection of abnormal traffic behavior such as running of red lights, driving in the wrong lane, or driving against traffic regulations.

Claims

1. An abnormal behavior detection system using an automatic classification of multiple features, comprising:

a feature extraction unit, used to extract features for one or more objects from monitoring data, and establishing feature sets of objects;

a behavior model establishing unit, receiving the feature set from the feature extraction unit;

then establishing and updating a behavioral model after a cluster analysis;

a behavior determining unit, receiving the feature set established by the feature extraction unit and the behavior model from the behavior model establishing unit, and further determining if there is an abnormal behavior; and

an output unit, used to output data of the behavior model and determination result of the object'"'"'s behavior.

2. The system of claim 1, wherein the feature extraction unit comprises a spatial feature extraction module and a temporal feature extraction module for respectively extracting spatial and temporal information.

3. The system of claim 2, wherein the system is applied to detect traffic violation, the spatial feature extraction module further comprises:

a traffic signal status detection subunit, used to obtain status of the traffic signal by an image analysis process; and

a vehicle position detection subunit, used to obtain one or more positions of one or more vehicles through the image analysis process.

4. The system of claim 2, wherein the system is applied to detection of the traffic violation, and the spatial feature extraction module further comprises:

a vehicle type detection subunit, used to detect the type of each vehicle by the image analysis techniques.

5. The system of claim 2, wherein the system is applied to detection of traffic violation, and the temporal feature extraction module further comprises:

a vehicle tracking subunit, obtaining position distribution of one or more vehicles along a continuous timeline, and obtaining the track of each vehicle.

6. The system of claim 1, wherein the monitoring data is received from each sensor distributed in the monitoring environment.

7. The system of claim 1, wherein the output unit is a storage device, a display, or a transmitter, or in combination of any number of the storage device, the display and the transmitter.

8. An abnormal behavior detection method which performs automatic classification based on multiple features, comprising:

extracting a feature set of each object from the monitoring data;

performing a feature cluster analysis based on similarity of all feature sets;

establishing or updating a behavior model based on the cluster analysis, and thereby defining the behavior model as a normal behavior model or an abnormal behavior, wherein the behavior model is formed by collecting the plurality of feature sets from the objects in a period of time;

comparing each object'"'"'s feature set with all behavior models, and determining whether the object fits the behavior model to identify the abnormal behavior, and

outputting the behavior type.

9. The method of claim 8, wherein the feature set of an object contains both spatial and temporal information.

10. The method of claim 8, wherein after the step of cluster analysis, the objects in the same group have similar environmental features, trajectory, and object features.

11. The method of claim 8, wherein the monitoring data is extracted from each sensor distributed in the monitoring environment.

12. The method of claim 11, wherein the monitoring data includes a road monitoring image and the status of traffic signal.

13. The method of claim 8, where in the step of establishing or updating the behavior model based on cluster analysis, a supervised learning scheme is introduced to define a specific behavior as an abnormal behavior.

14. The method of claim 8, where in the step of determining whether the object'"'"'s behavior fits any normal/abnormal behavior model;

the object'"'"'s behavior is regard as normal/abnormal, if the object'"'"'s feature set fits a normal/abnormal behavior model;

the object'"'"'s behavior is regard as abnormal, if the object'"'"'s feature set does not fit any normal behavior model.

15. The method of claim 8, where in the step of performing feature cluster analysis, all the objects are divided into several groups depend on the similarity of their feature set, and the behavior model of each group is established based on the probability distribution of the group which is obtained from the clustering results.