Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (7): 34-38.DOI: 10.3778/j.issn.1002-8331.2010.07.011

• 研究、探讨 • Previous Articles     Next Articles

Analysis on convergence of mean-shift and angle of continuous mean-shift vector

GUO Qing-chang,WANG Min-yi   

  1. CSIC-710 Research and Department Institute,Yichang,Hubei 443000,China
  • Received:2008-09-23 Revised:2008-12-22 Online:2010-03-01 Published:2010-03-01
  • Contact: GUO Qing-chang

均值移动算法收敛性及均值移动矢量夹角分析

郭庆昌,王敏毅   

  1. 中国船舶重工集团公司 第七一○研究所,湖北 宜昌 443000
  • 通讯作者: 郭庆昌

Abstract: Mean shift is an effective iterative algorithm.It has been widely used in the image processing and computer vision,but whose properties are not been perfectly proved.The convergence of the mean-shift is proved according to cauchy convergence criterion for the wrong and deficiency at the literatures.The angle of the mean shift vectors based on the arbitrary kernel in the same window is less than 90 degree.

Key words: mean-shift, convergence, Cauchy convergence criterion

摘要: 均值移动算法是一种统计迭代算法,目前在图像处理中得到了广泛应用。但是对其性质的分析仍然不完善,针对以往文献对均值移动算法收敛性证明的错误和不足,根据柯西收敛定理严格证明了均值移动算法的收敛性;证明了基于任意核,两连续均值移动矢量的夹角都不大于90°。

关键词: 均值移动, 收敛性, 柯西收敛定理

CLC Number: