【摘要】 针对传统的线性和非线性图像滤波算法对含有丰富细节的图像滤波处理的不足,提出一种基于模糊神经网络的图像滤波算法。该算法利用神经网络的自学习功能训练标准样本以确定模糊贴近度阈值,在基本不改变原有图像的灰度信息的前提下,找出图像受到噪声污染的像素点,采用迭代中值滤波算法得到的中值替换噪声点。仿真实验表明该方法有很好的滤波效果,优于传统的图像滤波算法。更多还原

【Abstract】 To absolve the faults of traditional linear and nonlinear image filter algorithms in processing images full of details,an image filter method based on fuzzy neural network is introduced in this paper.This method utilizes fuzzy similarity of fuzzy dimensional elements and study ability of neural network to train network.It can find the pixels polluted by noise,then replace those pixels by the results of median filter,while not change grey information of primary image.Experiments indicate this method can get better filter effect than traditional image filter methods.更多还原