计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (18): 13-17.DOI: 10.3778/j.issn.1002-8331.2009.18.004

• 博士论坛 • 上一篇    下一篇

粗神经网络及其在图像融合中的应用研究

朱颢东1,2,钟 勇1,2   

  1. 1.中国科学院 成都计算机应用研究所,成都 610041
    2.中国科学院 研究生院,北京 100039
  • 收稿日期:2009-02-20 修回日期:2009-03-27 出版日期:2009-06-21 发布日期:2009-06-21
  • 通讯作者: 朱颢东

Intellectualized system of image fusion based on rough neural network

ZHU Hao-dong1,2,ZHONG Yong1,2   

  1. 1.Chengdu Institute of Computer Application,Chinese Academy of Sciences,Chengdu 610041,China
    2.Graduate University of Chinese Academy of Sciences,Beijing 100039,China
  • Received:2009-02-20 Revised:2009-03-27 Online:2009-06-21 Published:2009-06-21
  • Contact: ZHU Hao-dong

摘要: 粗集理论能够优选数据,但容错性与推广能力比较弱;而神经网络具有较强的自组织、容错以及推理能力,却不能优选数据。把这两种理论结合起来,使之发挥各自优势,然后把它们用于图像融合之中,并提出了一种基于粗神经网络的图像融合方法,该方法使用遗传算法作为神经网络的训练算法。通过仿真实验表明,在对融合来自同一景物的多幅带噪声图像的应用中,该方法取得了很好的效果。

关键词: 粗集, 神经网络, 图像融合, 粗神经网络, 遗传算法

Abstract: Rough set theory can optimize data,but the fault-tolerant and promotion capabilities are relatively weak.Whereas neural network has strong self-organization and promotion capabilities,but can not optimize data.This paper combines rough set theory with neural network to exert their advantages,then applies them to image fusion and proposes a image fusion method based on rough neural network theory.The method uses a genetic algorithm to train the neural network.Simulation results show that,in the integration of features from the same number of images with the application of noise,the method is effective and feasible.

Key words: rough set, neural network, image fusion, rough neural network, genetic algorithm