J4 ›› 2010, Vol. 9 ›› Issue (4): 30-33.

• 数学与计算机科学 • 上一篇    下一篇

基于独立成分分析的自适应图像滤波算法

  

  1. 1.大理学院现代教育技术中心,云南大理 671003;2.大理学院数学与计算机学院,云南大理 671003
  • 收稿日期:2009-12-14 修回日期:2010-03-23 出版日期:2010-04-15 发布日期:2010-04-15
  • 作者简介:王蒙,助教,主要从事图像处理,模式识别,网络化控制研究.

Image Adaptive Filtering Algorithm Based on Independent Components Analysis

  1. 1.Modern Education Technology Center, Dali University, Dali,Yunnan 671003,China;
    2.College of Mathematics and Computer Science, Dali University, Dali, Yunnan 671003,China
  • Received:2009-12-14 Revised:2010-03-23 Online:2010-04-15 Published:2010-04-15

摘要:

近年来提出的独立成分分析(ICA)算法基于非高斯分布度量,可以有效提取图像边缘等重要信息。基于FastICA算法,利
用“MIT图像数据库”分别对不同场景图像集进行图像块采样,构建基于单一场景的训练集,训练各训练集得到ICA滤波器,得到并分析各组ICA滤波器对同类型图像集的滤波结果。实验表明ICA滤波可以实现不同场景图像的稀疏编码,且滤波结果具有对训练集自适应拟合能力。

关键词: 图像滤波, 自适应滤波, 独立成份分析, 稀疏编码

Abstract:

In recent years, we can pick up important information from images with the proposed independent components analysis (ICA) algorithm based on non-Gaussian distribution metric. In this paper , The ICA filters were obtained by MIT image dataset based on FastICA algorithm, by this algorithm , the sampling of image pitches from different scenes can be obtained by constructing a training sets based on one scene, and the results of each group of ICA filters were obtained and analyzed. The experiments indicated that ICA filtering can gain sparse code of different scenes images, and it has adaptive characters to training sets.

Key words: image filtering, adaptive filtering, independent components analysis,  sparse encode

中图分类号: