Method and system for separating illumination and reflectance using a log color space

• US 7,596,266 B2
• Filed: 04/13/2006
• Issued: 09/29/2009
• Est. Priority Date: 04/13/2006
• Status: Active Grant

• Current Assignee: Innovation Asset Collective
• Original Assignee: Tandent Vision Science, Inc.
• Inventors: Smith, Casey Arthur, Friedhoff, Richard Mark, Maxwell, Bruce Allen
• Primary Examiner(s): Do; Anh Hong
• Application Number:
  US11/403,719
• Publication Number:
  US 20070242878A1
• Time in Patent Office: 1,265 Days
• Field of Search: 382/156, 382/168, 382/162, 382/167, 382/163, 382/166, 348/222.1, 348/E9.01, 348/229.1, 345/589, 350/1.9, 235/469, 235/454, 235/462.41
• US Class Current: 382/167
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: H04N 1/6086 by scene illuminant, i.e. conditions at the time of picture capture, e.g. flash, optical filter used, evening, cloud, daylight, artificial lighting, white point measurement, colour temperature

Abstract

In a first exemplary embodiment of the present invention, an automated, computerized method for generating an illumination invariant, chromaticity representation of an image is provided. The method of the present invention comprises the steps of providing an image comprising an array of pixels, each pixel having N band color values, transforming the N bands to log color space values in a log color space, generating a bi-illuminant chromaticity plane in the log color space and projecting the log color space values to the chromaticity the plane to provide chromaticity representation values corresponding to the pixels of the image.

Claims

1. An automated computerized method for generating an illumination invariant, chromaticity representation of an image, comprising the steps of:

a computer executing the following steps;

providing an image comprising an array of pixels, each pixel having N band color values;

transforming the N bands to log color space values in a log color space;

generating a bi-illuminant chromaticity plane in the log color space; and

projecting the log color space values to the chromaticity plane to provide chromaticity representation values corresponding to the pixels of the image.

2. The method of claim 1, comprising the further step of mapping N band values of the pixels to the corresponding chromaticity representation values to generate a chromaticity representation of the image.

3. The method of claim of claim 1, wherein the N bands comprise RGB values for each pixel.

4. The method of claim 1, wherein the step of generating a bi-illuminant chromaticity plane in the log color space is carried out by estimating an orientation of an optimal normal to the plane, N.

5. The method of claim 4, wherein the step of estimating an orientation of an optimal normal to the plane, N is carried out by executing an entropy minimization technique.

6. The method of claim 5 wherein the entropy minimization technique is executed by executing a search algorithm for establishing a selection of orientations of the normal, N, for each selected orientation, estimating the entropy of the selected orientation, and selecting N corresponding to the orientation with a minimum estimated entropy.

7. The method of claim 6 wherein the search algorithm is selected from the group consisting of an exhaustive search algorithm, a univariate search algorithm, and a simulated annealing search algorithm.

8. The method of claim 6 wherein the step of estimating the entropy of the selected orientation is carried out by building a histogram of the chromaticity representation values, dividing the histogram into an array of bins, for each bin, calculating a probability of occupation by chromaticity representation values and calculating an entropy as a function of the probability of occupation, and summing the entropy calculations for all bins of the histogram.

9. The method of claim 5 wherein the entropy minimization technique is executed by executing a constrained search method.

10. The method of claim 9, wherein the step of executing a constrained search method is carried out by estimating entropy at each of an increment between bounds for a magnitude of a spectral ratio for the image, incrementing between the bounds by a preselected incremental amount.

11. The method of claim 10, wherein, at each increment of the magnitude, N is selected as N=log(1+1/(M Sn).

12. The method of claim 10 comprising the further step of iterating the step of incrementing, reducing the increment for a subsequent iteration.

13. The method of claim 4, wherein the step of estimating an orientation of an optimal normal to the plane, N is carried out by manually selecting lit and shadow portions of a material depicted in the image, and utilizing the selected portions to determine an estimate of the orientation.

14. The method of claim 13, wherein the step of utilizing the selected portions to determine an estimate of the orientation is carried out by estimating an orientation as a function of a log(average of lit area) and a log(average of shadow area).

15. The method of claim 4, wherein the step of estimating an orientation of an optimal normal to the plane, N is carried out by estimating the orientation as a function of a characteristic spectral ratio of the image.

16. The method of claim 15, wherein the step of estimating the orientation as a function of a characteristic spectral ratio of the image is carried out by executing N=log(1+1/S_vector), wherein S_vector is the characteristic spectral ratio.

17. An automated computerized method for generating an illumination invariant, material image, comprising the steps of:

a computer executing the following steps;

providing an original image comprising an array of pixels, each pixel having N band color values;

generating a histogram on a bi-illuminant chromaticity plane, the histogram being divided into bins;

calculating an envelope value for each bin of the histogram;

co-relating each of the pixels of the original image to a corresponding bin of the histogram; and

mapping the N band color values of each pixel to the envelope value of the co-related bin to provide the material image.

18. The method of claim 17 comprising the further step of utilizing the material image to generate an illumination image.

19. The method of claim 17 wherein the step of calculating an envelope value for each bin of the histogram is carried out by sorting pixels corresponding to the bin by intensity of the N band color values and selecting the envelope value as a function of the intensity of the N band color values.

20. The method of claim 19 wherein the step of selecting the envelope value as a function of the intensity of the N band color values is carried out by selecting a maximum value of the intensity as the envelope value.

21. The method of claim 19 wherein the step of selecting the envelope value as a function of the intensity of the N band color values is carried out by selecting a set percentile intensity.

22. The method of claim 19 wherein the step of selecting the envelope value as a function of the intensity of the N band color values is carried out by calculating the envelope value using isodata of the intensity of the N band color values.

23. The method of claim 22 wherein the step of calculating the envelope value using isodata of the intensity of the N band color values is carried out by executing an isodata algorithm to generate bright and dark clusters of values, and calculating a mean value for the values in the bright cluster as the envelope value.

24. The method of claim 19 wherein the step of selecting the envelope value as a function of the intensity of the N band color values is carried out by using a barbell model.

25. The method of claim 24 wherein the step of calculating the envelope value using a barbell model is carried out by generating a one dimensional histogram on a normal to the chromaticity plane based upon the intensity of the N band color values, initializing a barbell model using values from the one dimensional histogram and fitting the barbell model to the one dimensional histogram.

26. A computer system which comprises:

a CPU; and

a memory storing an image file defining an image comprising an array of pixels, each pixel having N band color values;

the CPU being arranged and configured to execute a routine;

to transform the N bands to log color space values in a log color space;

to generate a bi-illuminant chromaticity plane in the log color space; and

to project the log color space values to the chromaticity plane to provide chromaticity representation values corresponding to the pixels of the image.

27. A computer system which comprises:

a CPU; and

a memory storing an image file defining an image comprising an array of pixels, each pixel having N band color values;

the CPU being arranged and configured to execute a routine;

to generate a histogram on a bi-illuminant chromaticity plane, the histogram being divided into bins;

to calculate an envelope value for each bin of the histogram;

to co-relate each of the pixels of the original image to a corresponding bin of the histogram; and

to map the N band color values of each pixel to the envelope value of the co-related bin to provide a material image.

28. A computer program product, disposed on a computer readable media, the product including computer executable process steps operable to control a computer to:

provide an image comprising an array of pixels, each pixel having N band color values;

transform the N bands to log color space values in a log color space;

generate a bi-illuminant chromaticity plane in the log color space; and

project the log color space values to the chromaticity plane to provide chromaticity representation values corresponding to the pixels of the image.

29. A computer program product, disposed on a computer readable media, the product including computer executable process steps operable to control a computer to:

generate a histogram on a bi-illuminant chromaticity plane, the histogram being divided into bins;

calculate an envelope value for each bin of the histogram;

co-relate each of the pixels of the original image to a corresponding bin of the histogram; and

map the N band color values of each pixel to the envelope value of the co-related bin to provide a material image.