Abstract: Color quantization is a key technology in digital image processing. Based on the segmentation and clustering algorithm for color quantization, this article puts forward a kind of new color quantization algorithm upon K-means clustering. First of all, take a coarse classification to the pixels according to tonal H. Then, take quantization processing to each child clustering group based on k-means clustering color. Through the comparison with the current commonly used quantitative algorithm comes to a conclusion that this algorithm has a better performance in quantization effect and execution tmue.

Key words: Color quantization, HSI, K-means