CLASSIFICATION OF TARGET OBJECTS IN MOTION

• US 20120195459A1
• Filed: 01/28/2011
• Published: 08/02/2012
• Est. Priority Date: 01/28/2011
• Status: Active Grant

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Raytheon Company (Rtx Corporation)
• Inventors: Schmidt, Michael S., Prativadi, Prakruti S., Griffin, Shane A., Phelps, Ethan J.
• Granted Patent: US 8,625,905 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G01S 13/87 Combinations of radar systems, e.g. primary radar and secondary radar
• CPC Class Codes: G01S 7/003 Transmission of data between radar, sonar or lidar systems and remote stations
• CPC Class Codes: G01S 7/415 Identification of targets based on measurements of movement associated with the target
• CPC Class Codes: G06V 40/20 Movements or behaviour, e.g. gesture recognition recognition of facial expressions G06V40/16

Abstract

A method for classifying objects in motion that includes providing, to a processor, feature data for one or more classes of objects to be classified, wherein the feature data is indexed by object class, orientation, and sensor. The method also includes providing, to the processor, one or more representative models for characterizing one or more orientation motion profiles for the one or more classes of objects in motion. The method also include acquiring, via a processor, feature data for a target object in motion from multiple sensors and/or for multiple times and trajectory of the target object in motion to classify the target object based on the feature data, the one or more orientation motion profiles and the trajectory of the target object in motion.

Claims

1. A method for classifying objects in motion, comprising:

providing, to a processor, feature data for one or more classes of objects to be classified, wherein the feature data is indexed by object class, orientation, and sensor;

providing, to the processor, one or more representative models for characterizing one or more orientation motion profiles for the one or more classes of objects in motion; and

acquiring, via a processor, feature data for a target object in motion and trajectory of the target object in motion to classify the target object based on the feature data, the one or more orientation motion profiles and the trajectory of the target object in motion.

2. The method of claim 1, wherein providing the one or more representative models for characterizing one or more orientation motion profiles comprises acquiring orientation motion data for an exemplary object in motion.

3. The method of claim 1, wherein providing the one or more representative models for characterizing one or more orientation motion profiles comprises generating orientation motion data based on an analytical model for an exemplary object in motion.

4. The method of claim 1, comprising acquiring the feature data for the target object in motion and the trajectory of the target object in motion at one or more instances of time, periods of time, or a combination of both.

5. The method of claim 1, comprising acquiring the feature data for the target object in motion and the trajectory of the target object in motion using a plurality of sensors.

6. The method of claim 1, wherein the feature data comprises at least one of radar data, optical data or infrared data for each of the one or more classes of objects.

7. The method of claim 6, wherein the feature data comprises radar cross section signals and time derivatives of the radar cross section signals.

8. The method of claim 6, wherein the sensor is selected from the group consisting of a radar system, lidar system, optical imaging system, or infrared monitoring system.

9. The method of claim 1, comprising classifying the target object using Bayes'"'"' Rule the target object as belonging to a particular class of the one or more classes of objects based on the posterior probability the target object corresponds to the particular class.

10. The method of claim 1, wherein the feature data for the one or more classes of objects and the one or more representative models for characterizing one or more orientation motion profiles for the feature data are indexed in a database stored on the processor.

11. A system for classifying objects in motion, comprising:

data collected prior to classifying a target object in motion, the data comprising;

a) feature data on one or more classes of objects to be classified, wherein the feature data is indexed by orientation, sensor, and object class; and

b) one or more representative models for characterizing one or more orientation motion profiles for the feature data on the one or more classes of objects;

at least one sensor to acquire feature data for a target object in motion and trajectory of the target object in motion;

a first processor to generate reference feature data while the target object is in motion based on the object class and the trajectory of the target object in motion, wherein the at least one sensor provides feature data and time of feature data; and

a second processor to classify the target object during motion of the target object based on the reference feature data generated by the first processor, and feature data for the target object in motion.

12. The system of claim 11, comprising one or more sensors to acquire orientation motion data for an exemplary object in motion to generate the one or more orientation motion profiles for the one or more classes of objects.

13. The system of claim 11, wherein the first processor generates reference feature vectors by:

1. selecting an object class,

2. selecting an orientation motion profile for the selected object class,

3. for each point in time that feature data is collected by the at least one sensor for the target object in motion, the selected orientation motion profile and the trajectory of the target object in motion are used to determine the orientation of the target object in motion, a feature is selected from the feature database based on the sensor, object class, and orientation of the target object in motion.

14. The system of claim 13, wherein steps 1-3 are repeated to generate a collection of reference feature vectors.

15. The system of claim 11, wherein the second processor performs Bayesian classification using the reference feature data generated by the first processor as a priori data and the feature data for the target object to be classified to generate posterior object class type probabilities.

16. The method of claim 15, wherein the feature data for the target object in motion comprises feature data collected from each sensor at single points in time.