基于改进K-means的图片主色提取方法  

Image Color Extraction Algorithm Based on Improved K-means

作　　者：朱培恺 刘敬浩[1] Zhu Peikai;Liu Jinghao(School of Electronic Information Engineering,Tianjin University,Tianjin 300072,China)

机构地区：[1]天津大学电子信息工程学院

出　　处：《南开大学学报（自然科学版）》2019年第6期12-18,共7页Acta Scientiarum Naturalium Universitatis Nankaiensis

摘　　要：提出了一种基于密度的聚类结果优化方法,利用传统K-means快速高效的特性和基于密度的思想,有效提取图像中的主色,避免了基于密度来定位初始聚类中心导致的运算速度缓慢的缺点,同时避免传统聚类算法无法分析出部分主色的问题.实验结果表明,提出的方法能够很好地提升聚类效果,且最终得到的主色分析结果也更精确.An clustering results optimized method is proposed based on density.We utilize the characteristics of traditional K-means fast and efficient characteristic add density-based idea,effectively extract the dominant color in the image.It not only boost speed compared with K-means algorithm base on density,but also can extract the main color of the image.Experimental results show that the proposed method can enhance the effect of clustering,and the finally main color obtained results of the analysis are more accurate.

关 键 词：K-MEANS聚类 基于密度 欧氏距离 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]