Artificial ants and fireflies can perform colour quantisation

Abstract

Several methods based on the behaviour of biological systems have been proposed during the last years to solve a wide range of problems. Ant algorithms and Firefly algorithms are two methods of this type.The Ant-Tree for Colour Quantisation algorithm was recently proposed to quantise colour images. It is a clustering-based method which creates a tree structure where the pixels of the image are grouped in subtrees according to their similarity. A threshold value determines if the similarity is sufficient to associate a pixel to a subtree. The computation of that threshold includes a parameter in (0,1], whose value must be set taking into account the number of colours of the quantised image. Since this parameter influences the quality of the final image, it is recommended to perform several tests with different values of that parameter and then select the best result. In order to reduce the influence of said parameter and improve the resulting image, this paper proposes to combine the Ant-Tree for Colour Quantisation algorithm with the Firefly algorithm.Computational experiments show that the new method generates better images than some popular colour quantisation methods such as Median-cut, Octree, Neuquant, Wu’s method or the Variance-based method.