Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (3): 85-87.

• 学术探讨 • Previous Articles     Next Articles

Neuro fuzzy system-based image de-noising method

XU Hao   

  1. College of Information Science and Technology,Liaoning University,Shenyang 110036,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2008-01-21 Published:2008-01-21
  • Contact: XU Hao

一种基于模糊神经系统的图像去噪方法

徐 皓   

  1. 辽宁大学 信息科学与技术学院,沈阳 110036
  • 通讯作者: 徐 皓

Abstract: A method of de-noising a digital image contaminated by Gaussian noise is presented.This method can improve the performance of current Gaussian noise filters,reduce the blurring effect caused by filters.A Fuzzy Inference System(FIS) is properly designed and it is trained by ANFIS.The internal parameters of the FIS is adjusted and optimized through training.The training image data is generated automatically by computer programs.The optimized FIS can process the input data,produce enhanced images.The output images and the calculations show that this method can reduce blurring effects and improve the filter performance.Neuro fuzzy system can be used in image de-noising problems.If the membership functions,rules and training data are properly selected,this method can combine with other filters and enhance their performance.

Key words: neuro fuzzy system, Gaussian noise, filter, ANFIS

摘要: 提出一种对含有高斯噪声的数字图像的去噪方法,这种方法能够增强高斯噪声滤波器的性能,减少去噪对图像造成的模糊和失真。设计了一个模糊推理系统(FIS),并利用ANFIS训练这个FIS。通过训练可以调整、优化FIS的内部参数值。训练图像数据由计算机程序自动生成。优化后的FIS即可处理输入的图像数据,产生增强的图像。从结果图像的视觉效果和量化标准两方面的实验和分析,可以看出这种方法可基本消除高斯噪声滤波器产生的模糊和失真,提高滤波器性能。实验表明模糊神经系统可以应用于图像去噪问题。在合理地选择隶属度函数、规则和训练数据的前提下,会产生明显的图像增强效果。

关键词: 模糊神经系统, 高斯噪声, 滤波器, ANFIS