Technion engineer clears up underwater photos
Posted on 04.05.2004 - 14:57 EDT in SCIENCE & TECH NEWS by Rons_ROV_Links
A carefully constructed algorithm and a commonplace camera filter can not only dramatically
improve the quality of underwater photography, but may someday save lives as well.
The technique developed
over the past 18 months by scientists from the Technion-Israel Institute of Technology have started as a means of
improving underwater photography. But it has wide applications in engineering, science, and even lifesaving, and has
aroused much interest around the world – even before Dr. Yoav Schechner is due to present the work before the prestigious
IEEE Conference on Computer Vision and Pattern Recognition in Washington, DC, in July.
Schechner, of the
Technion's electrical engineering faculty, first intended to tackle a problem he encountered as an amateur diver.
"My professional field is optics and using computer vision to solve optical effects that cause visibility
degradation. I have dived from time to time in the Gulf of Eilat and off Australia, and was curious about dealing with the
problem of taking photos underwater," he told The Jerusalem Post.
"Even in the cleanest water, you have poor
visibility. There are many reasons for it. We analyzed the physical factors.
"The main problem," said Schechner,
"is the ambient light from the surroundings; it undergoes scattering into the line of sight which is called backscatter.
We and others understood that if the backscatter can be neutralized, the photo image would be much better."
Schechner and master's degree student Nir Karpel read "about 100 articles and books in the field" after
developing their hypothesis about using an algorithm and attaching a simple polarizing filter – which costs less than $100
and is widely used for conventional (overland) photography. Marine animals also use polarization for improved vision
"Our hunch was that if you take underwater photos with this filter and use a mathematical analysis of
physical things that occur in water, you can undo these distortions and compensate for them. At first we had no equipment
and had no real experience in this, so we took some scientific risk when we developed our hypothesis. No one in the world
had previously used this mathematical approach," Schechner added.
When they did get equipment and dived with it,
they "saw the algorithms didn't work so well, so we fine-tuned them until they did."
Their method, to be
outlined in an article that will be published in the IEEE conference's official proceedings, not only makes it possible
to see objects that previously appeared blurry and out of focus, but also makes it possible to estimate distances
underwater and give the photos three-dimensional depth. It would be a boon for marine biologists, Schechner said, because
photographers wouldn't have to disturb animals with closeups. The technique could also help lifesavers better see into
swimming pools to prevent drowning and might even be applied to improve photography in living tissue.
can be used for checking and monitoring underwater pipes and cables, bridge pylons, drilling rigs, the infrastructure of
artificial islands, archeological sites, robots working in water, sailing vessels, and atomic reactors. It also could be
of great assistance in scientific research (biology, underwater archeology, and underwater mapping) – and, of course, for
amateur underwater photographers.
03 May 2004