一种基于均匀颜色空间的色彩量化聚类算法

主要研究了将彩色图像重新量化为仅有几种颜色的色彩量化问题.将RGB非均匀颜色空间变换到CIE1976L*a*b*均匀颜色空间,并根据图像在L*a*b*颜色空间中的a*,b*取值范围来自动选取最初始的聚类中心,再通过K均值动态聚类算法和最近临准则得到最终的量化结果.实验结果表明,算法具有一定的实用性,量化效果也较为理想.

A clustering algorithm for color images quantization in a homogeneous color space

Some color quantization problems concerning re-quantization of a color image into a few numbers of colors are discussed.According to the value ranges of the color image's a*,b*,the firstly initial representative-points of the categories are automatically chosen in the homo-geneous color space of L*a*b*,which is transformed from the nonhomogeneous color space of RGB.Then,through the K-mean adaptive clustering and min-distances shortcut criterion the ultimate effects can be obtained.Experimental results show that this method is useful and the quantized effects can be accepted.