Color Quantization Based Artificial Bee Colony And Training Tools

Abstract

Color quantization addressed as a useful method to help with limited disk capacity, network bandwidth, monitors true resolution. It works in reducing the true colors to the colors that the human eyes can detect and don’t change the scene. In this paper a new method for color image quantization is proposed that depends on hybrid artificial tools. It consists of three main steps, the first is creating K colors to depend on them in quantizing the required colors in this step k-means algorithm is used starting from N colors. Then Artificial Bee Colony algorithm is used to improve the center of each cluster (color) which is combined to k-means algorithm as part of its optimization. Finally, the proposed color palette is used to quantize the image depending on the result of hybrid approach and apply to k-means. Two metrics are used to optimized the method (MSE, PSNR), both of them show good results for the quality of quantized images, as well as good human perception and good running time.