摘要
在服装图像处理与分析中,准确快速的图像分割是进行后续图像分析与理解工作的基础。在服装图像分割中为了克服K均值聚类算法的性能受初始聚类中心的选取影响的问题,提出了二次聚类的分割思想。先利用自适应的Mean Shift算法,得到区域数目n以及n个区域的类心向量,然后调用K均值算法进行二次聚类。与常用的分水岭分割算法的对比实验表明,提出的算法提高了原始K均值算法的时间效率与分割效果,最终的分割结果具有实用性。
In the clothing image processing and analysis,image segmentation is the basis of follow-up work.To overcoming the shortcomings of K-means algorithm of selecting original clustering center,this paper presents twice clustering ideal in the clothing image segmentation.At first mean shift algorithm is using to get region number of nand regional class center vector adaptively and then call the K-mean algorithm to conduct cluster.Experimental results show the algorithm proposed in this paper improves the initial clustering results on time efficiency and segmentation effect. The final segmentation results are practical.
出处
《电子测量技术》
2013年第8期53-55,60,共4页
Electronic Measurement Technology
关键词
服装
图像分割
均值移动
聚类
clothing
image segmentation
Mean Shift
cluster
