Algorithm AI Can Fix Grainy Images Without Looking at Clean Photos

Algorithm is a procedure or formula for solving a problem, based on conductiong a sequence of specified actions. A computer program can be viewed as an elaborate algorithm. In mathematics and computer science, an algorithm usually means a small procedure that solves a recurrent problem. Algorithms are widely used throughout all areas of IT.

Researchers from NVIDIA, Aalto University, and MIT have developed an AI algorithm, first of its kind, that can remove noise and artifacts from images without having to look at clean photos in advance.

In a research paper published last week, researchers say they trained their system with tens of thousands of other grainy photos.

This resulted in the creation of an AI algorithm that doesn’t need to know what a “clean” image looks like to reconstruct photos to their original state.

Algorithm

Researchers say their algorithm can be used to denoise old grainy photos, remove text-based watermarks, clear up medical MRI scans taken with undersampled inputs, enhancing astronomical photography, and denoise synthetically-generated (Monte Carlo) images.

Academics say that the training of such an algorithm takes similarly to what it would take to train an algorithm with clean images, providing no overhead for those interested in applying this method.

Grainy

Trained and Tested

To create their noise-filtering AI, the researchers started by adding noise to 50,000 pairs of clean images. The only difference between the two images in each pair was the pattern of noise added to them. The researchers then fed these pairs of grainy images to their AI, training it to remove the noise to reveal a cleaned-up version of the photo, one that looked almost identical to the image before the noise was added.

After training their AI, the researchers tested it using three sets of images to which they’d added noise. They found the system could de-noise the photos in milliseconds, producing a version just slightly softer-looking than the original before the researchers added noise.

Noise

Previous Nvidia AI work

This is not the first time NVIDIA’s research team has been behind impressive research. In April, NVIDIA and other researchers created another AI-based algorithm that can reconstruct images from which large chunks of content has been removed.

In May, NVIDIA and other fellow researchers created an AI algorithm that can teach robots to do various tasks just by observing a few repetitions by human workers.

NVIDIA