数字油画制作中的加速k均值颜色聚类算法

数字油画制作中的加速k均值颜色聚类算法 The Accelerated k-Means Algorithm Based on the Triangle Inequality for Clustering Color Images 苏清华 SU Qinghua 1 2 first-author 黄樟灿 HUANG Zhangcan 2 汪金水 WANG Jinshui 2 湖北工程学院数学与统计学院 湖北工程学院数学与统计学院 School of Mathematics and Statistics,Hubei Engineering University School of Mathematics and Statistics,Hubei Engineering University 武汉理工大学理学院 武汉理工大学理学院 School of Science,Wuhan University of Technology School of Science,Wuhan University of Technology 提出了一个数字油画制作中的加速k均值图像颜色聚类算法:AkMTI-CCI算法.该算法应用最远优先原则初始化聚类中心,消除了颜色聚类结果对初始中心的依赖性.利用三角不等式减少聚类过程中距离的计算量,提高了颜色聚类的速度.数值实验表明:AkMTI-CCI算法提高了颜色聚类的速度且改善了颜色聚类的效果. An accelerated k-Means algorithm for image color clustering in auto-generating digital oil painting,called AkMTI-CCI,is proposed.Centers are initialized according to the"furthest first"heuristic,so the last clustering result is the only one.And for the triangle inequalities are used to increase the number of distance computing,the clustering speed is higher.The experiments also show that AkMTI-CCI speeds the image color clustering,and make the effect with clustering better. 数字油画 颜色聚类 k均值聚类 三角不等式 digital oil painting color clustering k-means triangle inequality 10.14188/j.1671-8836.2014.02.015 TP391.41 国家自然科学基金资助项目（61070009）;中央高校基础科研业务费专项资金（2012-YB-19） 2014-02-01 2021-04-01 2 60