Image Denoising Method Based On Convolutional Neural Network And Its Application

There are many advanced methods for image processing including Probabilities,PDES and Variational method.Recently,deep learning is widely used in image processing,especially in image denoising,image classification and image restoration.Deep learning uses neural networks to learn useful feature representation directly from data.The task of the paper is to develop a neural network for one-dimensional piecewise-constant image denoising and understand how a neural network works in image denoising by analyzing the influence of each parameter on the network performance.Visualization has been used to understand the convolution neural network more deeply.At the end of the experiment,the convolution neural network we developed is used to nature image denoising,the effect of denoising is illustrated by a series of concrete examples.