Model-based halftoning of color images.

Area of Science:

• Computer Vision
• Image Processing
• Color Science

Background:

• Traditional color printing models often overlook printer-specific characteristics and human visual perception.
• Existing gray-scale model-based techniques provide a foundation for advanced color image reproduction.

Purpose of the Study:

• To introduce novel models for color printers to improve printed image quality.
• To develop model-based techniques that leverage printer and human visual system properties.
• To investigate the design of blue-noise screens using printer models.

Main Methods:

• Development of two model-based techniques: modified error diffusion (MED) and least-squares model-based (LSMB).
• Proposal of a printer model incorporating ink dot overlap and spectral absorption.
• Analysis of algorithm separability based on the relationship between RGB and CMY color spaces.

Main Results:

• MED and LSMB algorithms produce high spatial resolution and visually pleasing textures.
• The proposed printer model effectively accounts for ink dot overlap and spectral properties.
• Separable algorithms yield high-quality, device-dependent color images suitable for non-colorimetric applications.

Conclusions:

• Model-based techniques offer a robust approach to enhancing color printer output quality.
• The developed printer models are crucial for designing effective blue-noise screens.
• The algorithms are computationally simple, robust to registration errors, and suitable for applications prioritizing visual appeal over exact color reproduction.