摘要
针对图像的有序盲分离技术,提出一种基于粒子群优化的盲源抽取方法。该方法首先根据图像信号的高阶统计特性构造用于估计分离向量的目标函数,然后通过改进的粒子群算法优化该函数,获得最佳分离向量,并实现图像信号的逐次恢复。仿真实验结果表明,该方法不仅能依四阶累积量的绝对值降序地实现图像信号的盲分离,还能同时分离服从超高斯分布的语音信号和服从亚高斯分布的图像信号。
To separate images from a mixed signal one by one,a blind source separation of instantaneous mixtures is conducted by using particle swarm optimization algorithm.A goal function is constructed based on the higher order statistics properties of image signals,and the optimal extracted vectors are determined through optimizing the goal function with improved particle swarm optimization algorithm to separate image signals one by one.Simulation results show that not only can the method achieve blind separation of mixed image signals with a decreasing order of absolute value of kurtosis,but also that of the mixture of voice signals of super Gaussian distribution and image signal of sub-Gaussian distribution.
出处
《中国航海》
CSCD
北大核心
2013年第4期1-6,20,共7页
Navigation of China
基金
国家自然科学基金项目(51179074
61071038
51309116)
集美大学科研基金资助项目(ZQ2013001)
关键词
船舶、舰船工程
粒群优化算法
盲分离
图像信号
超高斯分布
亚高斯分布
ship, naval engineering
particle swarm optimization
blind source separation
image signal
sup-Gussian distribution
sub-Gussian distribution
作者简介
王荣杰（1981-）,男,福建晋江人,讲师,博士生,主要从事智能信息处理和电力电子电路故障诊断研究.E-mail：Roger811207@163.com.
