Filtering image impulse noise based on fuzzy rough neural network
基于模糊粗神经网络的图像脉冲噪声滤除

A FRNN(Fuzzy Rough Neural Network) was introduced to keep up trade-off between noise attenuation and detail preservation in image processing.The FRNN is composed of fuzzy neurons and indifferentiable rough neurons.GA(Genetic Algorithm) which integrate mountain climbing is applied to tune the weight of the FRNN.The local search ability for optimum solution is improved in the later period of the learning process.The result of simulation indicates that the FRNN which integrate the ability to deal with fuzzy information and rough set information has better performance than BP network and RNN(Rough Neural Network) in image fusion for filtering.The FRNN is a sort of hybrid intelligent neural network which has good performance.