A New Human Perceptional Color Quantization Algorithm

The goal of color quantization is to remarkably reducing the number of colors in an image without obviously degrading its visual effect. The study of this paper proposed a new human perceptional color quantization algorithm which minimizing the quantization distortion under the perceptually uniform Munsell NBS distance measurements. This method chooses the most frequently used colors as seeds with a limitation of the optimal minimal distance between seeds, which is computed using an iterative algorithm. Experiments show that the proposed method outperforms the minimal variance quantization (MVQ) algorithm on quantization distortion and perceptual effect.