Optimizing Template for Lookup-Table Inverse Halftoning using Elitist Genetic Algorithm

Abstract

A template optimization method based on elitist genetic algorithm was proposed for lookup-table inverse halftoning. A mathematical model with constraint conditions was built to describe the template optimization problem. We solved this optimization problem by using elitist genetic algorithm and designed the details about encoding and decoding scheme, selection and reproduce, crossover, mutation, elitist strategy and fitness function according to the proposed optimization model. In experiments, we demonstrated the performance on Floyd-Steinberg error diffusion, Jarvis-Judice error diffusion, cluster dither, Bayer disperse dither and dot diffusion halftone images. According to our experiment study, our method approaches to the optimal result closer than the greedy algorithm and simulated annealing do. We suggested L = 10 on cluster dither images but L = 5 on other four kinds of halftone images.