Image restoration technology beyond scattering media (Image provided by KAIST).
DAEJEON, Sept. 1(Korea Bizwire) — Scientists at the Korea Advanced Institute of Science and Technology (KAIST) have unveiled an artificial intelligence technique that can reconstruct clear video images obscured by fog, smoke, frosted glass, or even human tissue.
The joint research team, led by Professor Jang Moo-seok of KAIST’s Department of Bio and Brain Engineering and Professor Ye Jong-chul of the Kim Jaechul Graduate School of AI, said Sunday it developed a video diffusion–based restoration model that overcomes distortions caused by what scientists call “scattering media.” Such materials scramble the paths of light, rendering visual information blurred or indecipherable.
The newly developed system learns the temporal correlations in video frames, allowing it to restore images even when the scattering environment shifts over time — such as landscapes seen through a curtain swaying in the wind. Unlike earlier approaches limited by training datasets, the model can adapt to dynamic conditions, the researchers said.
Potential applications are wide-ranging: noninvasive medical diagnostics capable of visualizing beneath skin or blood, improved visibility in fire rescues filled with smoke, and safer driving on fog-covered roads.
The findings were published August 13 in the international journal IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
Kevin Lee (kevinlee@koreabizwire.com)
