A fast and novel technique for color quantization using reduction of color space dimensionality

摘要: Abstract This paper describes a fast and novel technique for color quantization using reduction of space dimensionality. The histogram is repeatedly sub-divided into smaller classes. colors each class are projected on carefully selected line, such that the dis-similarities preserved. Instead principal axis class, line defined by mean vector largest distance away from color. composed projection values then used to cluster two representative palette colors. As result, computation in process fast. A pixel mapping algorithm based proposed data clustering also presented this paper. Experimental results show algorithms quantize images with high image quality efficiently.