Image segregation system architecture

• US 20100142825A1
• Filed: 12/05/2008
• Published: 06/10/2010
• Est. Priority Date: 12/05/2008
• Status: Active Grant

• Current Assignee: Innovation Asset Collective
• Original Assignee: Tandent Vision Science, Inc.
• Inventors: Kwak, Iljung Samuel, Leonardus Nabbe, Bartholomeus Christianus, Diebel, Milan, Rodgers, Timothy King JR., Smith, Casey Arthur, Tolliver, David Allen, Bushell, Steven Joseph, Friedhoff, Richard Mark, Stein, Andrew Neil, Bagnell, James Andrew, Maxwell, Bruce Allen, Dana, Kristin Jean
• Granted Patent: US 8,139,867 B2
• Time in Patent Office: Days
• US Class Current: 382/199
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/22 Source localisation; Inverse modelling

Abstract

In an exemplary embodiment of the present invention, an automated, computerized method is provided for processing an image. According to a feature of the present invention, the method comprises the steps of generating spatio-spectral information for the image, defining a constraint as a function of the spatio-spectral information, and performing an optimization operation as a function of the constraint to generate an intrinsic image corresponding to the image.

Claims

1. An automated, computerized method for processing an image, comprising the steps of:

providing an image;

generating spatio-spectral operators for the image;

defining a constraint between the spatio-spectral operators; and

performing an optimization operation as a function of the constraint to generate an intrinsic image corresponding to the image.

2. The method of claim 1 wherein the intrinsic image comprises one of a material image and an illumination image.

3. The method of claim 1 wherein the spatio-spectral operators comprise operators selected from the group consisting of tokens, token map information, log chromaticity representation values, X-junctions, BIDR model representations, textons, and combinations thereof.

4. The method of claim 1 wherein the constraint comprises a constraint selected from the group consisting of an anchor constraint, a same illumination constraint, a smooth illumination constraint, a Type B token constraint, a Linear token constraint, a BIDR enforcement constraint, a same texton histogram constraint, a log chromaticity similarity constraint, an X junction constraint, and combinations thereof.

5. The method of claim 1 wherein the step of performing an optimization operation is carried out by performing a convex optimization operation.

6. The method of claim 1 wherein the step of performing an optimization operation includes the further step of introducing factors established as a function of real world illumination and material phenomena, to keep material/illumination values determined in the optimization operation, to within physically plausible limits.

7. The method of claim 6, wherein the factors include bounds.

8. An automated, computerized method for processing an image, comprising the steps of:

providing an image;

organizing spatio-spectral information for the image in a matrix equation expressed by;

[A] [x]=[b], wherein [A] expresses values determined by a constraining relationship imposed upon the spatio-spectral information, [b] expresses recorded information for the image, and [x] expresses an unknown material/illumination component of the image;

performing an optimization operation to determine a solution for the equation; and

utilizing the solution to generate an intrinsic image corresponding to the image.

9. The method of claim 8 wherein the step of performing an optimization operation is carried out by performing an optimization operation selected from the group consisting of a least squares solver and a preconditioned conjugate gradient solver.

10. The method of claim 8 wherein the step of performing an optimization operation includes the further step of introducing factors established as a function of real world illumination and material phenomena, to keep material/illumination values determined in the optimization operation, to within physically plausible limits.

11. The method of claim 10, wherein the factors include bounds.

12. The method of claim 11, wherein the step of performing an optimization operation is carried out by performing an optimization operation selected from the group consisting of a limit 1, limit infinity solver, a bounded least squares solver and a bounded limit 1, limit infinity solver.

13. The method of claim 8 wherein the intrinsic image comprises one of a material image and an illumination image.

14. A computer system which comprises:

a CPU; and

a memory storing an image file containing an image;

the CPU arranged and configured to execute a routine to perform a routine to generate spatio-spectral operators for the image, define a constraint between the spatio-spectral operators, and

perform an optimization operation as a function of the constraint to generate an intrinsic image corresponding to the image.

15. A computer system which comprises:

a CPU; and

a memory storing an image file containing an image;

the CPU arranged and configured to execute a routine to perform a routine to organize spatio-spectral information for the image in a matrix equation expressed by;

[A] [x]=[b], wherein [A] expresses values determined by a constraining relationship imposed upon the spatio-spectral information, [b] expresses recorded information for the image, and [x] expresses an unknown material/illumination component of the image, perform an optimization operation to determine a solution for the equation, and utilize the solution to generate an intrinsic image corresponding to the image.

16. A computer system comprising:

a memory storing an image file containing an image;

a constraint generator library storing a plurality of constraint software modules, each one of the constraint software modules defining spatio-spectral information utilized by the respective one of the constraint software modules and an expression of a constraining relationship imposed upon the respective spatio-spectral information;

a service provider; and

a solver;

wherein;

the service provider operates to select a constraint software module from the plurality of constraint software modules, arrange spatio-spectral information relevant to the image as a function of the selected constraint software module, and input the arranged spatio-spectral information relevant to the image to the solver, the solver operates to perform an optimization operation as a function of the arranged spatio-spectral information relevant to the image, to generate an intrinsic image corresponding to the image.

17. The computer system of claim 16, further comprising an operator cache for storing spatio-spectral information defined by the plurality of constraint software modules.

18. The computer system of claim 16 further comprising a GUI for a user input;

the service provider operates to select a constraint software module as a function of the user input.

19. The computer system of claim 16 wherein the arranged spatio-spectral information relevant to the image comprises a matrix equation expressed by:

[A] [x]=[b], wherein [A] expresses values determined by a constraining relationship imposed upon the spatio-spectral information, [b] expresses recorded information for the image, and [x] expresses an unknown material/illumination component of the image.

20. An automated, computerized method for processing an image, comprising the steps of:

providing an image;

generating spatio-spectral information for the image;

defining a constraint as a function of the spatio-spectral information; and

performing an optimization operation as a function of the constraint to generate an intrinsic image corresponding to the image.

21. The method of claim 20 wherein the constraint comprises a constraint selected from the group consisting of an absolute material color constraint, a same material constraint and a relative reflectance constraint.

22. The method of claim 20 wherein the spatio-spectral information for the image comprises an operator selected from the group consisting of tokens, token map information, log chromaticity representation values, X-junctions, BIDR model representations, textons, and combinations thereof.

23. A computer system which comprises:

a CPU; and

a memory storing an image file containing an image;

the CPU arranged and configured to execute a routine to generate spatio-spectral information for the image, define a constraint as a function of the spatio-spectral information, and perform an optimization operation as a function of the constraint to generate an intrinsic image corresponding to the image.

24. An automated, computerized method for processing an image, comprising the steps of:

providing an image;

generating spatio-spectral information for the image;

defining a set of constraints as a function of the spatio-spectral information; and

performing an optimization operation as a function of the constraints, wherein individual ones of the set of constraints are weighted relative to one another, in the optimization operation, according to preselected factors, to generate an intrinsic image corresponding to the image.

25. The method of claim 24 wherein the preselected factors are selected from a group consisting of perceived importance of the respective ones of the constraints, strength, an empirically determined confidence level and combinations thereof.