改进的FastICA算法在图像盲分离中的应用

摘要

An improved FastICA algorithm is presented based on basic principle and fast algorithm of independent component analysis (ICA). A new and simple object function which synthesize kurtosis and skewness is developed,Then propose an improved flow of FastICA algorithm. The improved algorithm is apllied to blind image separation, The experimental result shows the improved algorithm enhance the effect of independent component separation.%在系统研究独立分量分析基本原理、快速算法的基础上,提出了一种改进的独立分量分析快速算法.引入了一种综合考虑峰度和偏度的新的简单目标函数,并对独立分量分析的快速算法流程提出了改进.改进的算法在图像盲分离中得到了应用.实验结果表明,该算法显著提高了独立分量分离效果.