how to test nosiy images in common scenarios ?

[hufangjian]

Hi， Thank you for your efforts.
but I have question.
In your codes:
seqn = seq + noise
your input are clean data and noise which generate by (" noise = torch.empty_like(seq).normal_(mean=0, std=args['noise_sigma']).to(device)")

But common scenarios ,I got some images with noise. i dont not know the noise distribution.

how can remove image noise and get the clean images.
just like a sequences of these images

[m-tassano]

Hi,
If I understand correctly, you want to denoise photographic noise, not only Gaussian.
In this case, you can denoise your sequences with the model provided. You will probably have to test with different values for the noise map to see the one that works the best.
Else, you can retrain the model. In this case, you can choose to remove the noise map completely (blind denoising). A second option would be to use a noise estimator, and use its output as noise map.
Hope this helps

[Scienceseb]

Hi,
If I understand correctly, you want to denoise photographic noise, not only Gaussian.
In this case, you can denoise your sequences with the model provided. You will probably have to test with different values for the noise map to see the one that works the best.
Else, you can retrain the model. In this case, you can choose to remove the noise map completely (blind denoising). A second option would be to use a noise estimator, and use its output as noise map.
Hope this helps

Hello @m-tassano, I am wondering what kind of noise estimator (network) will you used to get a noise map on real noisy image ? I'm working on fluorescent microscopy I have sequence of noisy images and also the sequence of the pseudo non-noisy images, but I dont have the noise map, cause it is a real noise. Thanls

[m-tassano]

Hi @Scienceseb , I have not implemented that for fastdvdnet, but I would probably start testing the one proposed with CBDNet. Hope this helps