top of page

Underwater Image Reconstruction

​

Underwater images suffer from color distortion and low contrast, because light is attenuated while it propagates through water. Attenuation underwater varies with wavelength, unlike terrestrial images where attenuation is assumed to be spectrally uniform. The current underwater image formation model descends from atmospheric dehazing equations where attenuation is a weak function of wavelength. We showed that this model introduces significant errors and dependencies in the estimation of the direct transmission signal. We proposed a revised equation for underwater image formation. 

​

The revised physically accurate model showed: 

  1. the attenuation coefficient of the signal is not uniform across the scene but depends on object range and reflectance

  2. the coefficient governing the increase in backscatter with distance differs from the signal attenuation coefficient.

​

More recently, we developed Sea-thru, an algorithm that recovers color with our revised model using RGBD images. The Sea-thru method estimates backscatter using the dark pixels and their known range information. Then, it uses an estimate of the spatially varying illuminant to obtain the range-dependent attenuation coefficient.

 

We also address the problem of underwater single-image color restoration using haze-lines. The attenuation depends both on the water body and the 3D structure of the scene, making color restoration difficult.  By estimating just two additional global parameters: the attenuation ratios of the blue-red and blue-green color channels, the problem is reduced to single image dehazing, where all color channels have the same attenuation coefficients. 

 

Also, we aim to constrain the set of physically-feasible wideband attenuation coefficients in the ocean by utilizing water attenuation measured worldwide by oceanographers. We calculate the space of valid wideband effective attenuation coefficients in the 3D RGB domain and find that a bound manifold in 3-space sufficiently represents the variation from the clearest to murkiest waters. Finally, we suggest a method for estimating the medium properties (both attenuation and scattering) using only images of backscattered light from the system’s light sources.

​

Datasets

T_S04859.jpg
depthT_S04859.jpg
T_S04887.jpg
depthT_S04887.jpg

SEA-THRU DATASET

Dataset includes 5 subsets with RAW images and corresponding depth maps. 

katzaa_LFT_4019.jpg
katzaa_LFT_4024.jpg
katzaa_LFT_4034.jpg
katzaa_LFT_4031.jpg

SQUID: STEREO QUANTITATIVE UNDERWATER IMAGE DATASET

The database contains 57 stereo pairs from four different sites in Israel, two in the Red Sea (representing tropical water) and two in the Mediterranean Sea (temperate water). The dataset includes RAW images, TIF files, camera clibration files, and distance maps.

Publications

2020UnveilingOpticalProperties_Bekerman_

Unveiling Optical Properties in Underwater Images. Yael Bekerman, et al., 2020.

2018color-restorationa.jpg

Underwater Single Image Color Restoration Using Haze-Lines and a New Quantitative Dataset. Dana Berman, et al., 2018.

2017attenuationcoefficients.png

What is the Space of Attenuation Coefficients in Underwater Computer Vision? Derya Akkaynak and Tali Treibitz, 2017.

2020ColorRestorationD.jpg

Underwater Single Image Color Restoration Using Haze-Lines and a New Quantitative Dataset. Dana Berman, et al., 2020.

2018revised-underwater-imagec.jpg

A Revised Image Formation Model. Derya Akkaynak and Tali Treibitz, 2018.

2017in-situCalibrationb.jpg

In Situ Target-Less Calibration of Turbid Media. Ori Spier, et al., 2017.

2019seathrua.png

Sea-Thru: A Method for Removing Water from Underwater Images. Derya Akkaynak and Tali Treibitz, 2019.

2017divingIntoHazeLinesb.jpg

Diving into Haze-Lines: Color Restoration of Underwater Images. Dana Berman, et al., 2017.

bottom of page