A Noise Filtering Method of Adaptive Fuzzy Threshold for Wavelet

针对小波变换多分辨分析(MRA)的特点,本文提出一种多尺度分级的自适应模糊权重中值滤波的去噪方法.首先,利用开关控制策略的模糊理论建立隶属函数,用高斯自适应模型对噪声点进行预检测,然后在每一级小波变换过程中应用自适应模糊中值滤波(AFWMF)算法进行噪声滤波.实验表明,常规的小波去噪方法只能去除图像中的高斯噪声,该方法既能去除高斯噪声也能去除非高斯噪声.与中值滤波方法相比,该方法在去噪的同时能保留大量的原图像边缘、细节等重要信息,具有更好的去噪效果.According to the feature of the MRA,this paper presents a noise filtering method of adaptive fuzzy threshold for wavelet classification base on multi resolution.First,subject function is brought forward according to switch controlling of fuzzy strategy.Adaptive gauss model is established Ior pre-detecting.Finally,the adaptive fuzzy weight media filter(AFWMF) is applied in each layer of wavelet transform.This algorithm needn’t know the uncontaminated signal as well as the problem of wavelet coefficient.The experimental results show that usual wavelet algorithm only eliminates gauss noise.However,our algorithm removes both gauss noise and non-gauss noise.Compared with media noise filtering,it preserves original image ingormation of edge and details while removes noise.国家自然科学基金(60175008)资