摘要
图像分色在纺织和印刷等行业中有着广泛而重要的应用,其目的是用尽量少的颜色来描述一幅彩色图像,使得到的分色图像与原图像尽可能的接近。提出一种基于单遍聚类和K-均值聚类相结合的自适应图像分色算法。该算法首先对原图像颜色进行统计学习,由单遍聚类产生初始调色板,然后根据该调色板对原图像的像素点进行K-均值聚类,产生分色图像。实验结果表明,与单纯K-均值聚类算法相比,该算法能在提高分色图像质量的同时进一步减少颜色数。
Color image separation is widely and importantly applied in textile and printing industry, its aim is to use as less kinds of color as possible to describe a piece of true color image and make it much similar to original image. An adaptive color separation algorithm combined of Onee- clustering and K- means clustering is presented. This method firstly does the statistic learning of the original image, create the initial color palette by Onee - clustering, than clusters the pixels in the original image by K - means according to color palette, then obtains the separated image. The result of the experiment shows that, comparing with simple K - means clustering algorithm, this method could further minish the amount of color at the time of improving the color separation image quality.
出处
《计算技术与自动化》
2006年第1期110-113,共4页
Computing Technology and Automation
基金
宁波大红鹰职业技术学院软件学院基金资助(I0603)
关键词
单遍聚类
K-均值聚类
分色
调色板
Once -clustering
K means clustering
color,separation
color palette
