Why do you want to ask next DnCNN can realize blind denoising, ordinary CNN can realize blind denoising? Just started to learn, there are many don't understand the question,

CodePudding user response:

1, DnCNN realize blind denoising key lies in the training set, he said in a blind denoising training set adopted ranging in [0, 50] intensity of noise, such training can make the model has a certain robustness, moreover its blind denoising using the model of the number of layers and the size of the convolution kernels also compared with precise denoising made certain adjustment;
2, common network denoising can implement blind is arguably denoising, but need to constantly adjust parameters to achieve the best effect, training full is also very important, of course, please

CodePudding user response:

1, blind denoising and open-label denoising is whether know the noise level
2, now do not blind representatives of denoising is DNCNN, belongs to the representative of the gaussian noise blind denoising image denoising is cbdnet as it really
In 3, DNCNN DNCNN - b blind denoising refers to under the precondition of gaussian noise, do not know the noise level, do not know whether the blind denoising problem of blind denoising are consistent with you