•  
  •  
 

Journal of Undergraduate Research

Keywords

dithering, palette selection, image quantization, colors

College

Physical and Mathematical Sciences

Department

Computer Science

Abstract

When performing image quantization, palette selection has always been a difficult problem. All methods, although they may produce acceptable result for some images, perform poorly on others. The problem lies in deciding which colors are best to include in the quantized palette. For example, when quantizing an image that has a large gradient sky and multiple colors, a computer scientist is faced with a difficult choice. Either he can choose to allocate the colors of the palette to have a wide range of colors or he can give most of the colors to the gradient areas. The first will produce a brighter and more colorful image but leave the color gradients (e.g., sky) with few colors and a grainy look. The latter will produce beautiful gradients but the image will tend to be dull and washed out. We provide an algorithm that creates a balance in quantized palettes between color gradients and unique colors.

Download

Share

COinS