A Hybrid Approach Using the k‐means and Genetic Algorithms for Image Color Quantization
Marcos Roberto e Souza, Anderson Carlos Sousa e Santos, Hélio Pedrini · 2020
This chapter proposes and evaluates a hybrid method that combines k-means and genetic algorithm for image color quantization. The genetic algorithm aims to improve the quantized images generated by the k-means algorithm through a qualitative objective function, which considers the local neighborhood. In a general fashion, color quantization algorithms can be categorized as uniform and adaptive. A color quantization based on the radius weighted mean cut was proposed to construct a color palette. The chapter discusses an algorithm that presents the pseudocode of the proposed color quantization method. Extensive experiments were conducted on different images to demonstrate the effectiveness of the proposed hybrid method. The results demonstrated that there is a consistent improvement in terms of structural similarity index when compared to the outcomes obtained with the k-means algorithm. The chapter intends to explore other evolutionary techniques in the context of image color quantization.