计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (28): 199-201.DOI: 10.3778/j.issn.1002-8331.2009.28.060

• 图形、图像、模式识别 • 上一篇    下一篇

使用新的码字分割方法的快速LBG算法

刘 刚1,刘 晶2,王 泉1   

  1. 1.西安电子科技大学 计算机外部设备研究所,西安 710071
    2.西安理工大学 计算机科学与工程学院,西安 710048
  • 收稿日期:2008-05-26 修回日期:2008-09-26 出版日期:2009-10-01 发布日期:2009-10-01
  • 通讯作者: 刘 刚

Fast LBG algorithm using new codeword splitting method

LIU Gang1,LIU Jing2,WANG Quan1   

  1. 1.Research Institute of Computer Peripherals,Xidian University,Xi’an 710071,China
    2.School of Computer Science and Engineering,Xi’an University of Technology,Xi’an 710048,China
  • Received:2008-05-26 Revised:2008-09-26 Online:2009-10-01 Published:2009-10-01
  • Contact: LIU Gang

摘要: 针对LBG算法中初始码书生成存在盲目性的问题,提出了一种基于训练向量集合分量之间的相关性进行码向量分割的方法。在迭代过程,记录每一个码向量所映射的训练向量子集;生成新的初始码书时,首先计算被分割码向量所映射的训练子集的第一个分量与其他各分量之间的协方差;根据协方差来分割相应的码向量。实验表明,应用新的码向量分割机制后,LBG算法中的迭代次数和运行时间减少了约10%左右。

关键词: LBG算法, 初始码书, 码向量, 协方差

Abstract: To alleviate the blindness of Linde-Buzo-Gray algorithm(LBG algorithm) while generating initial codebook,a new codeword splitting method based on covariance of training vector is proposed.Subset of training vector mapped to each codeword is recorded in the iteration step.In the codeword splitting step,covariance between column one and other columns of each subset is calculated and the corresponding codeword is split into twos according to the covariance.Experiment results show that the number of iteration and runtime of LBG algorithm are reduced about 10% by applying the new codeword splitting mechanism.

Key words: LBG algorithm, initial codebook, codeword, covariance

中图分类号: