Intuitively it seems like this "reduces" noise by the same mechanism that blurring "reduces" noise -- by using averaged samples we are in effect low-passing the image, which filters out the high-frequency noise.
BenPust
I saw this machine learning/statistical approach to noise reduction a while ago (https://arxiv.org/pdf/1803.04189.pdf) that uses a neural network that is trained on images that are modified to have noise that way the network can reduce noise without blurring the image.
zsano1
Is there any difference between downsampling and subsampling?
michaeltu1
Downsampling refers to decreasing the number of pixels in an image. In order to accomplish this, we choose an interpolation method - i.e. averaging, bicubic, or subsampling. Subsampling is one way to do downsampling.
Intuitively it seems like this "reduces" noise by the same mechanism that blurring "reduces" noise -- by using averaged samples we are in effect low-passing the image, which filters out the high-frequency noise.
I saw this machine learning/statistical approach to noise reduction a while ago (https://arxiv.org/pdf/1803.04189.pdf) that uses a neural network that is trained on images that are modified to have noise that way the network can reduce noise without blurring the image.
Is there any difference between downsampling and subsampling?
Downsampling refers to decreasing the number of pixels in an image. In order to accomplish this, we choose an interpolation method - i.e. averaging, bicubic, or subsampling. Subsampling is one way to do downsampling.