Multipass data integration for automatic detection and classification of objects

• US 8,275,172 B2
• Filed: 10/30/2009
• Issued: 09/25/2012
• Est. Priority Date: 10/30/2009
• Status: Active Grant

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Raytheon Applied Signal Technology Incorporated (Rtx Corporation)
• Inventors: Wilby, Andrew D., Harbaugh, J. Kent
• Primary Examiner(s): Wu, Jingge
• Assistant Examiner(s): James, Thomas A
• Application Number:
  US12/610,243
• Publication Number:
  US 20110103642A1
• Time in Patent Office: 1,061 Days
• Field of Search: None
• US Class Current: 382/103
• CPC Class Codes: G06F 18/253 of extracted features
• CPC Class Codes: G06V 2201/07 Target detection

Abstract

Classification of a potential target is accomplished by receiving image information, detecting a potential target within the image information and determining a plurality of features forming a feature set associated with the potential target. The location of the potential target is compared with a detection database to determine if it is close to an element in the detection database. If not, a single-pass classifier receives a potential target'"'"'s feature set, classifies the potential target, and transmits the location, feature set and classification to the detection database. If it is close, a fused multi-pass feature determiner determines fused multi-pass features of the potential target and a multi-pass classifier receives the potential target'"'"'s feature set and fused multi-pass features, classifies the potential target, and transmits its location, feature set, fused multi-pass features and classification to the detection database.

Claims

1. An apparatus comprising:

an input port configured to receive first pass image information and second pass image information;

a detector module configured to detect a potential target within the first pass image information and the second pass image information, and determine a plurality of features forming a feature set associated with the potential target and a new location of the potential target within the second pass image information, wherein detection is based upon one or more of at least some of the determined features and a location of the potential target within the first pass image information;

a detection database configured to store at least a location, features, fused multi-pass features and a classification of the potential target;

a comparator module configured to compare the new location of the potential target with contents of the detection database to determine whether the new location of the potential target is within a predetermined range of the location of an element in the detection database;

a single-pass classifier module configured to (a) receive the potential target'"'"'s feature set, if the new location of the potential target is not within said predetermined range of the location of an element in the detection database, (b) classify the potential target according to received potential target'"'"'s feature set, and (c) transmit the location, the feature set and the classification of the potential target to the detection database;

a fused multi-pass feature determiner module configured to determine fused multi-pass features of the potential target by combining and comparing the features sets of the potential target corresponding to the new location of the potential target, if the new location of the potential target is within said predetermined range of the location of an element in the detection database; and

a multi-pass classifier module configured to (a) receive the potential target'"'"'s feature set and said determined fused multi-pass features, if the new location of the potential target is within said predetermined range of the location of an element in the detection database, (b) classify the potential target according to the received potential target'"'"'s feature set and fused multi-pass features, and (c) transmit the location, the feature set, the fused multi-pass features and the classification of the potential target to the detection database.

4. The apparatus of claim 1, wherein the detector module is configured to determine said plurality of features using image segmentation.

5. The apparatus of claim 1, wherein the potential target classification is stored in the detection database.

6. The apparatus of claim 1, further comprising a training system to generate training parameters needed by the single pass and multi-pass classifiers, using the detection database.

7. The apparatus of claim 1, wherein the first pass and second pass image information are received from one or more sensors on a moving platform.

8. The apparatus of claim 1, wherein the potential target is one or more of the group consisting of manganese nodules, ship wrecks, rocks on the sea floor, images from a satellite or a reconnaissance aircraft, and suspect items in radiographic medical images.

2. A method, comprising:

receiving first pass image information;

detecting a potential target within the first pass image information and determining a plurality of features forming a feature set associated with the potential target, wherein the detecting is based upon one or more of at least some of the determined features and a location of the potential target;

storing said determined plurality of features in a detection database;

receiving second pass image information;

detecting a new location of the potential target within the second pass image information;

comparing the new location of the potential target with contents of the detection database to determine whether the new location of the potential target is within a predetermined range of the location of an element in the detection database;

using a single-pass classifier configured to (a) receive the potential target'"'"'s feature set, if the potential target is not within said predetermined range of the location of an element in the detection database, (b) classify the potential target according to received potential target'"'"'s feature set, and (c) transmit the location, the feature set and the classification of the potential target to the detection database;

combining and comparing the feature sets of the potential target corresponding to the new location of the potential target to determine fused multi-pass features of the potential target if the new location of the potential target is within said predetermined range of the location of an element in the detection database; and

using a multi-pass classifier configured to (a) receive the potential target'"'"'s feature set and said determined fused multi-pass features, if the new location of the potential target is within said predetermined range of the location of an element in the detection database, (b) classify the potential target according to the received potential target'"'"'s feature set and fused multi-pass features, and (c) transmit the location, the feature set, the fused multi-pass features and the classification of the potential target to the detection database.

9. The method of claim 2, wherein said determining the plurality of features comprises using image segmentation.

10. The method of claim 2, further comprising storing the potential target classification in the detection database.

11. The method of claim 2, further comprising generating training parameters needed by the single pass and multi-pass classifiers by a training system, using the detection database.

12. The method of claim 2, wherein the first pass and second pass image information are received from one or more sensors on a moving platform.

13. The method of claim 2, wherein the potential target is one or more of the group consisting of manganese nodules, ship wrecks, rocks on the sea floor, images from a satellite or a reconnaissance aircraft, and suspect items in radiographic medical images.

3. An apparatus, comprising:

means for receiving first pass image information second pass image information;

means for detecting a potential target within the first pass image information and second pass image information, and determining a plurality of features forming a feature set associated with the potential target and a new location of the potential target within the second pass image information, wherein the detecting is based upon one or more at least some of the determined features and a location of the potential target;

detection storage means configured to store at least a location, features, fused multi-pass features and a classification of the potential target;

means for comparing the new location of potential target with contents of the detection database means to determine whether the location of the potential target is within a predetermined range of the location of an element in the detection storage means;

mean for (a) receiving the potential target'"'"'s feature set, if the new location of the potential target is not close to the location of an element in the detection database means, (b) classifying the potential target according to received potential target'"'"'s feature set, and (c) transmitting the location, the feature set and the classification of the potential target to the detection storage means;

means for determining fused multi-pass features of the potential target, if the potential target is within a predetermined range of the location of an element in the detection storage means; and

means for (a) receiving the potential target'"'"'s feature set and said determined fused multi-pass features, if the new location of the potential target is within said predetermined range of the location of an element in the detection database (b) classifying the potential target according to the received potential target'"'"'s feature set and fused multi-pass features, and (c) transmitting the location,the feature set, the fused multi-pass features and the classification of the potential target to the detection storage means.

14. A method for detection and classification of objects, the method comprising:

detecting a potential target within a first pass image information obtained via scanning an area of interest, by determining a plurality of first features and a first location of the potential target;

determining whether said area of interest has been scanned prior to obtaining the first pass image information

when said area of interest has been scanned prior to obtaining the first pass image information, comparing the first location of the potential target with an old location of the potential target detected prior to obtaining the first pass image information to determine whether the new location is within a predetermined range of said location;

extracting a plurality of first features for the potential target at the first location, when the first location is within a predetermined range of old location;

combining the first features with old features extracted from the potential target detected prior to obtaining the first pass image information to classify the potential target; and

when the first location is not within a predetermined range of said old location, using said old features to classify the potential target.

15. The method of claim 14, wherein said determining the first plurality of features comprises using image segmentation.

16. The method of claim 14, further comprising storing the potential target classification in the detection database.

17. The method of claim 14, further comprising generating training parameters needed by the single pass and multi-pass classifiers by a training system, using the detection database.

18. The method of claim 14, wherein the first pass and prior pass image information are received from one or more sensors on a moving platform.

19. The method of claim 14, wherein the potential target is one or more of the group consisting of manganese nodules, ship wrecks, rocks on the sea floor, images from a satellite or a reconnaissance aircraft, and suspect items in radiographic medical images.