Abstract
In this paper, a vector quantization method for image coding inspired from the self-assembling behaviour observed in real ants where ants progressively become attached to an existing support and then successively to other attached ants is used is presented. This approach has been previously used to solve the clustering problem. The artificial ants are defined which quantize the groups of wavelet coefficients after building trees. The paper presents some of the constraints of the recently proposed Ant Colony System Vector Quantization (ACSVQ) algorithm and gets rid of them by using the proposed algorithm. It has been shown that the quantization of zero tree vectors using ant trees outerforms the traditionall used Linde-Buzo-Gray (LBG) algorithm and recently proposed ACSVQ coding technique.