A group of scientists from the University of Illinois, in preparation for the International Conference on Computer Vision and Pattern Recognition, presented an impressive video of how artificial intelligence helps to combat noise in photographs.
The authors of the study made two frames of each plot – strongly underexposed (in appearance black) and with correct exposure (without noise) due to long exposure – “reference”. In the end, artificial intelligence tried to bring the result of underexposed cadres to “reference” ones.
Examples are the usual methods and methods of AI:
Below is a comparison with the innovative X-Trans matrix and Fuji’s loudspeaker:
For those who want to learn more about the work, I recommend to refer to the PDF , in which the authors describe their studies in detail.