US 11,810,272 B2
Image dehazing and restoration
Avital Treibitz, Beer Yaacov (IL); Dana Berman, Yehud (IL); and Shai Avidan, Ra'anana (IL)
Assigned to CARMEL HAIFA UNIVERSITY ECONOMIC CORPORATION LTD., Haifa (IL); and RAMOT AT TEL-AVIV UNIVERSITY LTD., Tel Aviv (IL)
Filed by CARMEL HAIFA UNIVERSITY ECONOMIC CORPORATION LTD., Haifa (IL); and RAMOT AT TEL-AVIV UNIVERSITY LTD., Tel Aviv (IL)
Filed on Dec. 31, 2020, as Appl. No. 17/139,056.
Application 17/139,056 is a continuation of application No. 16/092,053, granted, now 10,885,611, issued on Dec. 31, 2020, previously published as PCT/IL2017/050426, filed on Apr. 6, 2017.
Claims priority of provisional application 62/319,338, filed on Apr. 7, 2016.
Prior Publication US 2021/0201452 A1, Jul. 1, 2021
Int. Cl. G06K 9/00 (2022.01); G06T 5/00 (2006.01); G06F 18/22 (2023.01); G06T 7/90 (2017.01); G06F 16/901 (2019.01)
CPC G06T 5/003 (2013.01) [G06F 16/9027 (2019.01); G06F 18/22 (2023.01); G06T 7/90 (2017.01); G06T 2207/10024 (2013.01); G06T 2207/20061 (2013.01)] 14 Claims
OG exemplary drawing
 
1. A method for dehazing a digital image, comprising:
operating at least one hardware processor to:
convert the digital image to one or more medium-compensated images that are each based on attenuation coefficient ratios of a different water type, wherein the digital image is of an underwater scene;
cluster hazy Red-Green-Blue (RGB) values of pixels of a digital image into haze-lines in an RGB space, wherein each of the haze-lines is comprised of a sub-group of the hazy RGB values that are scattered non-locally over the digital image;
estimate, based on the haze-lines, a transmission map of the digital image, wherein the transmission map encodes scene depth information for each pixel of the digital image; and
calculate a dehazed digital image based on the transmission map, by recovering true RGB values,
wherein performing the clustering, the estimating, and the calculating is for each of the one or more medium-compensated images.