[1]范文兵,邢军阳,李海涛,等.基于QPSO和ICA的图像盲分离方法研究[J].郑州大学学报(工学版),2012,33(03):106-109,112.[doi:10.3969/j.issn.1671-6833.2012.03.027]
　FAN Wenbing,XING Junyang,LI Haitao,et al.Research of lmage Blind Separation Method Based on QPSO and ICA[J].Journal of Zhengzhou University (Engineering Science),2012,33(03):106-109,112.[doi:10.3969/j.issn.1671-6833.2012.03.027]
点击复制

基于QPSO和ICA的图像盲分离方法研究()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
33
期数:
2012年03期
页码:
106-109,112
栏目:
出版日期:
2012-05-10

文章信息/Info

Title:
Research of lmage Blind Separation Method Based on QPSO and ICA
作者:
范文兵; 邢军阳; 李海涛; 等.
郑州大学 信息工程学院,河南郑州,450001, 郑州大学 信息工程学院,河南郑州,450001, 郑州大学 信息工程学院,河南郑州,450001, 郑州大学 信息工程学院,河南郑州,450001
Author(s):
FAN Wenbing; XING Junyang; LI Haitao; etc;
School of information Engineering, Zhengzhou University, Zhengzhou 450001 , China
关键词:
独立分量分析;  量子粒子群;  盲源分离;  负熵
Keywords:
independent component analysis;  quantum particle swarm optimization;  blind source separation; negentropy
分类号:
TN911.7
DOI:
10.3969/j.issn.1671-6833.2012.03.027
摘要:
针对ICA技术中常用的普通梯度算法容易陷入局部最优,提出了一种基于量子行为的粒子群算法和独立分量分析相结合的盲源分离新算法.以负熵作为独立分量分析的目标函数,用QPSO算法代替普通梯度算法,对瞬时混合信号进行分离,给出了算法的具体步骤.实验结果表明,该算法能够有效实现图像的盲源分离.同时与其他算法对比,体现了该算法更高的性能.
Abstract:
In this paper, we introduce the independent Component Analysis ( ICA) and Quantum ParticleSwarm Optimization ( PS0 ) brielly. As the ordinary gradient algorithm of ICA technology is easy to fall into loeal optimum , we proposed quantum-behavior based particle swarm optimization and independent component a.nalysis for blind source separation combining new algorithms. This algorithm takes negative entropy as the ob.jeetive function of independent component analysis, replaces the ordinary gradient algorithm with QPSO algorithm and separates the instantaneous mixed signals, All the steps of this algorithm are given in this paper.Experiment is show that the proposed algorithm can effectively achieve the image of the blind source separation, Compared with other algorithms, this algorithm shows belter performance.
更新日期/Last Update: 1900-01-01