Genetic algorithm with intelligent crossover for colour quantization

Abstract

Genetic algorithms (GAs) with crossover using heuristics can rapidly provide satisfying high-quality solutions to colour quantization problems that are known to be NP-complete. This paper proposes an intelligent genetic algorithm based colour quantization (IGACQ) algorithm. The crossover operation of the intelligent genetic algorithm (IGA) consists of economically identifying good individual genes from parents and intelligently combining these good genes to generate high-quality offspring. The merit of intelligent crossover without using heuristics is that the conventional random recombination and generate-and-test search for offspring are replaced with a divide-and-conquer strategy and a systematic reasoning recombination based on orthogonal experimental design. It is shown empirically that IGACQ performs better than existing GA-based and non-GA-based methods for colour quantization in terms of quantization quality.

Keywords:

• colour quantization
• clustering
• genetic algorithm
• intelligent crossover
• partitioning