Generating an image profile LUT for a N-color marking device via multi-objective cost minimization

1. A method for constructing a profile LUT for accurate color reproduction in a color marking device, the method comprising:

selecting at least one image quality attribute desired to be improved for a color marking device, each of said image quality attributes having an associated cost of applying said attribute to said device;

selecting a reference ICC profile to determine reference reflectance spectrum values for each node of a LUT of said color marking device;

for each node of said LUT:

finding candidate recipes which reproduce target L*a*b* colors associated with this node to within a color matching error which is less than a predefined threshold error e m ;

pruning said candidate recipes to a subset containing only those recipes where a spectral difference between a predicted spectra corresponding to N color levels for a given recipe for this node and a reference reflectance spectra corresponding to a spectra of a recipe produced by said selected reference ICC profile for target L*a*b* colors associated with this node, is less than a threshold distance d m , each of said candidate recipes having a cost determined with respect to each of said selected image quality attributes; and

identifying a candidate recipe from said subset of recipes that minimizes a multi-objective cumulative cost function J that is based upon a cost associated with each recipe in said subset of recipes relative to said selected image quality attributes, said multi-objective cost function comprises:

where ω i is the i th weight and ƒ i is a cost function of applying recipe X to the i th image quality attribute, ω n+1 is a weight applied to a distance in which X i,nbr is a recipe of the i th neighbor node of the current node, and α i is a weight applied to the i th recipe determined in a reverse proportion to a distance that neighbor node is from node V, calculated in an input color space of said LUT; and

generating a new profile LUT for said color marking device, said new profile LUT comprising, for each node, the candidate recipe which minimized that node's respective multi-objective cost function.

2. The method of claim 1 , wherein said node of said LUT starts from nodes along a neutral axis and proceeds outward towards the gamut boundary.

3. The method of claim 1 , wherein said selected image quality attribute comprises any of: visual color match, graininess, mottle, color stability, metamerism, ink cost, color difference, and spectral match.

4. The method of claim 1 , wherein said color recipes comprises using color model, spectra model, and image quality attributes models which relate color recipe values to image color, spectra, and image quality attributes.

5. A system for constructing a profile LUT for accurate color reproduction in a color marking device, the system comprising:

a memory and a storage device; and

a processor in communication with said memory and storage device, said processor executing machine readable instructions for performing:

selecting at least one image quality attribute desired to be improved for a color marking device, each of said image quality attributes having an associated cost of applying said attribute to said device;

selecting a reference ICC profile to determine reference reflectance spectrum values for each node of a LUT of said color marking device;

for each node of said LUT:

finding candidate recipes which reproduce target L*a*b* colors associated with this node to within a color matching error which is less than a predefined threshold error e m ;

pruning said candidate recipes to a subset containing only those recipes where a spectral difference between a predicted spectra corresponding to N color levels for a given recipe for this node and a reference reflectance spectra corresponding to a spectra of a recipe produced by said selected reference ICC profile for target L*a*b* colors associated with this node, is less than a threshold distance d m , each of said candidate recipes having a cost determined with respect to each of said selected image quality attributes; and

identifying a candidate recipe from said subset of recipes that minimizes a multi-objective cumulative cost function J that is based upon a cost associated with each recipe in said subset of recipes relative to said selected image quality attributes, said multi-objective cost function comprises:

where ω i is the i th weight and ƒ i is a cost function of applying recipe X to the i th image quality attribute, ω n+1 is a weight applied to a distance in which X i,nbr is a recipe of the i th neighbor node of the current node, and α i , is a weight applied to the i th recipe determined in a reverse proportion to a distance that neighbor node is from node V, calculated in an input color space of said LUT; and

generating a new profile LUT for said color marking device, said new profile LUT comprising, for each node, the candidate recipe which minimized that node's respective multi-objective cost function.

6. The system of claim 5 , wherein said node of said LUT starts from nodes along a neutral axis and proceeds outward towards the gamut boundary.

7. The system of claim 5 , wherein said selected image quality attribute comprises any of: visual color match, graininess, mottle, color stability, metamerism, ink cost, color difference, and spectral match.

8. The system of claim 5 , wherein said color recipes comprises using color model, spectra model, and image quality attributes models which relate color recipe values to image color, spectra, and image quality attributes.