Method and apparatus for correction of color cast in a digital image

• US 6,104,830 A
• Filed: 04/06/1998
• Issued: 08/15/2000
• Est. Priority Date: 06/22/1995
• Status: Expired due to Fees

• Current Assignee: ESKO-Graphics A/S (Danaher Corporation)
• Original Assignee: Purup-Eskofot A/S
• Inventors: Schistad, Bernhard
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Novik, Dmitry A.
• Application Number:
  US08/981,528
• Time in Patent Office: 862 Days
• Field of Search: 282/162, 282/167, 282/276, 282/281, 358/518, 358/520, 358/522
• 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: H04N 1/60 Colour correction or control H04N1/54 takes precedence
• CPC Class Codes: H04N 1/6077 Colour balance, e.g. colour cast correction
• 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

The invention relates to a method and an apparatus for the analysis and correction of colour casts in a digitally represented image. According to a preferred embodiment, the crominance contents of the image may be transformed by use of a Hough-transformation or the like transformation, wherein the line structures in the carthesic system of co-ordinates for the crominance content is transformed to a representation in a Hough-diagram in the form of a polar system of co-ordinates, since such line structure is hereby transformed into a point structure. The Hough-diagram is examined with a view to finding a cell in the diagram which has the highest value and the cell found is saved. subsequently cells are neutralized in the Hough-diagram within a predetermined angular distance, and the Hough-diagram is re-examined with a view to finding that diagram cell which, following neutralization, has the highest value. The cell found is saved and the cells saved which represent the maximae in the Hough-diagram are used to determine the intersecting points for lines in the crominance content system, and said intersecting points are used to evaluate the discolouration of the image.

Claims

1. A method of analyzing and optionally correcting color casts or coloring errors in a digitally represented image, wherein each photograph is represented by a number of pixels or photographic points in a color space with separate chrominance and luminance dimensions, wherein

a) part of or all of the chrominance dimensions of the color space are examined to establish completely or partially hidden line of convergence structures,

b) that two or more line-of-convergence structures are determined as parameterized functions, and that

c. on the basis of these line-of-convergence structures at least one mutual intersecting point is established for determining a color convergence of the digitally represented image;

and

wherein the examination of a Cartesian chrominance dimension of the color space comprises an approximation of a portion of or all of the image points to a parameterized function via a Hough-transformation; and

wherein the image points in the color space are approximated to a straight line in a polar system of co-ordinates via a modified Hough-transformation according to the formula

space="preserve" listing-type="equation">x cos θ

+y sin θ

=ρ

H.sub.log (a*,b*))

wherein x and y designate the Cartesian representation in a plane of a color space, θ

designates the angle to the axis of abscissa, ρ

is the angular distance from origin to the line, and wherein ##EQU4## the U(a*,b*) being a unit function which is summed up throughout the entire x,y-plane, wherein parameters a* and b* are the chrominance coordinates of the color space.

2. A method according to claim 1, wherein the color space is divided into color cells which consist of the quantized axes of chrominance and luminance.

3. A method according to claim 1, wherein at least two coincidental intersections are determined which are subsequently used for sorting out non-coincidental intersections.

4. A method according to claim 1, wherein the angle between the line-of-convergence structures for determining the intersection or the intersections in a chrominance space exceeds 10°

.

5. A method according to claim 1, wherein a weighting of the image points present in the chrominance dimension.

6. A method according to claim 1, wherein the examination of a Cartesian chrominance dimension of the color space comprises an approximation of a portion of or all of the image points to a parameterized function via a Hough-transformation.

7. A method according to claim 6, wherein the Hough-transformation approximates a portion of or all of a Cartesian chrominance representation (x,y) of all the photographic points to a polar representation (θ

,ρ

) as the polar representation is quantized in θ

,ρ

in such a manner that a line representation is depicted in an accumulator cell which consists of the respective θ

,ρ

-quantized co-ordinates.

8. A method according to claim 6, wherein the image points in the color space are approximated to a straight line in a polar system of co-ordinates via a Hough-transformation according to the formula

9. A method according to claim 6, wherein the intersections (x,y) of the line-of-convergence structures are found as expressed by:

space="preserve" listing-type="equation">x=(ρ

--y sin θ

)/cos θ

, and

space="preserve" listing-type="equation">y=(ρ

1 cos θ

2-ρ

2 cos θ

1)/sin(θ

1-θ

2), wherein

11. A method according to claim 1, wherein one or more line candidates are selected in a chrominance dimension whereupon a summation of the square of the distance of pixels to a line candidate is carried out, the line candidate being selected from among the pixels where the summation assumes a minimum.

12. A method according to claim 1, wherein the data of the digitally represented image are subsequently corrected depending on intersection ratios in such a manner that the intersecting points of the line structures are shifted to or near to a neutral point of a color specter or a color plane.

13. A method according to claim 1, wherein the data of the digitally represented image are subsequently corrected linearly depending on intersecting ratios in such a manner that the intersecting point of the line structures is displaced to or near to the neutral point of the color specter or the color plane.

14. A method according claim 1, wherein the color space is subdivided into a number of luminance layers, and that one or more luminance layers are analyzed separately.

15. A method according to claim 13, wherein each luminance layer is corrected separately if correction is possible.

17. An apparatus according to claim 15, wherein the computer comprises means for controlling a color application of a printing unit into a medium.

18. An apparatus according to claim 15, wherein the computer comprises means for controlling a color representation on a computer or television screen.

10. ρ

1. θ

1 represent the coordinates of the one line in the polar representation in the form of a point or an accumulator cell in the polar Hough-diagram, ρ

2, θ

2 represent the co-ordinates of the second line, and ρ

,θ

is any one of the coordinates of the two lines.

16. An apparatus for the correction of color casts in a digitally represented image according to a method comprising:

a method of analyzing and optionally correcting color casts or coloring errors in a digitally represented photographic image represented by a number of pixels or photographic points in a color space with separate chrominance and luminance dimensions, wherein

a) part of or all of the chrominance dimensions of the color space are examined to establish completely or partially hidden line of convergence structures,

b) that two or more line-of-convergence structures are determined as parameterized functions, and that

c. on the basis of these line-of-convergence structures at least one mutual intersecting point is established for determining a color convergence of the digitally represented image; and

wherein the examination of the Cartesian chrominance dimension of the color space comprises an approximation of a portion of or all of the image points to a parameterized function via a Hough-transformation; and

wherein the image points in the color space are approximated to a straight line in a polar system of co-ordinates via a modified Hough-transformation according to the formula

space="preserve" listing-type="equation">x cos θ

+y sin θ

=ρ

H.sub.log (a*,b*),

wherein x and y designate the Cartesian representation in a plane of a color space, θ

designates the angle to the axis of abscissa, ρ

is the angular distance from origin to the line and wherein ##EQU5## the U(a*,b*) being a unit function which is summed up throughout the entire x,y-plane, wherein parameters a* and b* are the chrominance coordinates of the color space and wherein the method is implemented in a computer for the processing of digital image data.