基于邻域灰度值聚类的图像色彩量化

在将图像中的多种颜色或灰度量化成数目较少的颜色或灰度的过程中,存在着计算过于复杂、量化后图像偏差较大等问题,鉴于此,提出基于邻域灰度值聚类的图像色彩量化方法.首先结合邻域像素的灰度和空间信息对像素进行一维灰度化;然后采用基于像素灰度加权系数的改进模糊 均值聚类算法对像素进行聚类.分析和实验表明,该方法可以减少量化计算的复杂度,保持图像的整体层次,量化后图像偏差较小,对图像处理具有一定的实用价值.

Image color quantization based on clustering of neighborhood gray level

For the main problems appearing in the process of reducing the number of either colors or gray level of the image,
for example, the computation tends to be too complex and the image deviation tends to be too big, the method of image
color quantization based on the clustering of neighborhood gray level is proposed. Firstly, the one-dimensional gray level
expression of pixels is carried out by integrating the space and the gray information of the neighborhood pixels. Then the
pixels are clustered by using the improved fuzzy ??-means clustering algorithm. The theoretical analysis and the experiment
show that the proposed method can reduce the complexity of the quantization computation and maintain the overall level
and the partial details of the image, and the image deviation becomes smaller after the quantization, which has the certain
practical value to the imagery processing.