Reducing the size of a high resolution profile lookup table

A system for generating a destination profile LUT to improve quality of color reproduction, the system comprising:

• a memory and a storage medium for storing data;
• and
• a processor in communication with said memory and said storage medium said processor executing machine readable program instructions to perform a method comprising:
• receiving a high resolution LUT comprising a high density cube on a structured grid in a color space of an image output device;
• selecting a size of a destination profile LUT comprising a high density cube on a structured grid in a color space of an image output device, said size defining a target number of nodes along a neutral axis of said destination profile LUT, said target number of nodes being less than a number of nodes on the neutral axis of said high density cube;
• repeating until all the subsets of nodes are considered:
• selecting a subset of nodes along a neutral axis of said high density cube, a number of nodes in said subset being equal to said target number of nodes inclusive of nodes at a start and endpoint of said neutral axis, said selecting effectuating a down-sampling of said high resolution LUT and producing a low resolution LUT;
• up-sampling said low resolution LUT to a desired number of nodes, said up-sampling producing an up-sampled reconstructed LUT;
• interpolating between each of said selected subset of nodes;
• calculating an error between color values between high resolution LUT and low resolution LUT;
• and
• selecting a next subset of nodes for a next iteration, at least a portion of nodes selected on a next iteration being different than nodes selected on any previous iteration;
• selecting the subset of nodes with minimum error;
• and
• generating a destination profile LUT for said image output device from said selected subset of nodes.

A computer implemented method for generating a destination profile LUT to improve quality of color reproduction, the method comprising:

• receiving a high resolution LUT Thigh comprising a high density cube on a structured grid of at least a 70-cube;
• identifying at least one region of high curvature of a color gamut of an image output device;
• repeating until all the subsets of nodes are considered:
• using a dynamic optimization technique to select a subset of nodes from said high density cube, said selecting effectuating a down-sampling of said high resolution LUT and producing a low resolution LUT Tlow not greater than 33-cube;
• up-sampling said low resolution LUT to a desired number of nodes, said up-sampling producing a reconstructed LUT, Treconstruct;
• calculating an error between color values of said reconstructed LUT and color values of said high resolution LUT;
• and
• selecting, for a next iteration, a next subset of nodes from said high density cube, said next subset of nodes being selected such that at least a portion of said next subset of nodes is different than the subset of nodes selected on any previous iteration;
• selecting the subset of nodes with minimum error;
• generating a destination profile LUT for said image output device from said selected subset of nodes with the minimum error;
• and
• using said destination profile LUT to profile said image output device.