Method and system for learning spatio-spectral features in an image

• US 20080075371A1
• Filed: 09/27/2006
• Published: 03/27/2008
• Est. Priority Date: 09/27/2006
• Status: Active Grant

• Current Assignee: Innovation Asset Collective
• Original Assignee: Tandent Vision Science, Inc.
• Inventors: Smith, Casey Arthur, Friedhoff, Richard Mark, Maxwell, Bruce Allen, Dana, Kristin Jean
• Granted Patent: US 7,747,079 B2
• Time in Patent Office: Days
• US Class Current: 382/199
• CPC Class Codes: G06F 18/211 Selection of the most significant subset of features
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/60 relating to illumination properties, e.g. using a reflectance or lighting model
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G06V 10/778 Active pattern-learning, e.g. online learning of image or video features

Abstract

In a first exemplary embodiment of the present invention, an automated, computerized method is provided for determining illumination flux in an image. According to a feature of the present invention, the method comprises the steps of performing a computer learning technique to determine spatio-spectral information for the images, and utilizing the spatio-spectral information to identify illumination flux.

Claims

1. An automated, computerized method for determining illumination flux in an image, comprising the steps of:

performing a computer learning technique to determine spatio-spectral information for the image;

and

utilizing the spatio-spectral information to identify illumination flux.

2. The method of claim 1 wherein the step of performing a computer learning technique to determine spatio-spectral information for the image is carried out by using a training set.

3. The method of claim 2 wherein the training set comprises positive examples and negative examples.

4. The method of claim 3 wherein the positive examples comprise synthetic examples of spatio-spectral features.

5. The method of claim 1 wherein the spatio-spectral information comprises an X-junction.

6. The method of claim 5 comprising the further step of using the X-junction to calculate a characteristic spectral ratio for the image.

7. The method of claim 6 comprising the further step of using the characteristic spectral ratio to identify illumination flux in the image.

8. The method of claim 7 wherein the illumination flux comprises an illumination boundary.

9. The method of claim 1 wherein the spatio-spectral information comprises a characteristic spectral ratio.

10. The method of claim 9 comprising the further step of using the characteristic spectral ratio to identify illumination flux in the image.

11. The method of claim 10 wherein the illumination flux comprises an illumination boundary.

12. The method of claim 3 wherein the training set comprises a feature vector set, the set including a feature vector for each positive example of the training set, each feature vector defined by characteristics of the spatio-spectral information.

13. The method of claim 12 comprising the further step of building a classifier as a function of the feature vector set.

14. The method of claim 13 comprising the further step of using the classifier to identify spatio-spectral information.

15. The method of claim 13 wherein the classifier comprises an Adaboost classifier.

16. The method of claim 12 wherein the characteristics are selected by a Probabilistic Learning Tree.

17. A computer system which comprises:

a CPU; and

a memory storing an image file;

the CPU arranged and configured to execute a routine to perform a computer learning technique to determine spatio-spectral information for the image file, and utilize the spatio-spectral information to identify illumination flux.