Articles Blog

This researcher created an algorithm that removes the water from underwater images

This researcher created an algorithm that removes the water from underwater images


I really see this as the start of the Artificial Intelligence boom in marine science. My name is Derya Akkaynak. And I am an oceanographer and engineer I specialize in problems of imaging and vision underwater. I am now in the Lembeh strait in Indonesia to test a new algorithm that I developed called Sea-thru, which takes an underwater image and removes the water from it, and it looks just like it would as if you took the photo
on land. Underwater images typically have
an overwhelming color cast, green or blue, depending on where you took them. Objects in far distances are occluded by
a layer of what we call backscatter, but think of it as a layer of haze. So, the further you are from the objects in
the scene the more haze you get in the scene. Because light as it travels through the water, gets absorbed and scattered, colors fade away. That’s where underwater images look so dull
and distorted all the time. [Music up] I’m diving with a regular consumer
camera, and I carry a color chart with me. Every time I see a reef with large 3D structure,
I place my color chart at the base of the reef, and then I swim away about 15 meters. Then I start swimming towards the reef, towards
the color chart, and photograph it from slightly different angles until I get to the reef and
then I swim over the reef, photograph the top of the reef, all sides. Once I have the distance information
all I do is I take the raw photos back to my computer, and I apply the method that uses
a certain mathematical formula, goes through each pixel and calculates what the degradation should be and removes it. Currently, in the field of underwater
imaging and marine science, we are at a standstill So if you’re a biologist, for example,
studying corals on the seafloor, and you made a survey of a reef, and you’d like to see
what species or what composition of animals are found on the seafloor, for the most part,
you have to do that manually, because the water takes away so much that you can’t see the true color of the species you’re looking for. This method is not photoshopping
an image. It’s not enhancing or pumping up the colors
in an image. It’s a physically accurate correction, rather
than a visually pleasing modification. I imagine in addition to scientists,
recreational divers or underwater photographers would also be very interested in using this
method, because finally from their images, they can remove all the degrading components
and see the vivid colors of a scene just the way they would have as if that scene was on
land. [credits]

96 thoughts on “This researcher created an algorithm that removes the water from underwater images”

  1. Reminds me of the person who invented an algorithm to remove clothes from people in photos but decided to yank it from the market when he realized that it would be misused.

  2. Seems rather easy, surprised it's not already a thing. Take color grading in picture , calibrate all other images according to the true colors. Done.

  3. Wait, she did what??? That's just incredible, thank you so much for sharing!!!

    Road to 1mil reached within 24 hours 👊😎

  4. [Undersea, with a device in hand looking at a reef about 16 ft above the ocean floor.]

    Researcher: "and with a push of a button, the water in the scene disappears"

    [all of the water LITERALLY DISAPPEARS, she falls directly on the coral, clearly in a bad way]

    Balding Groundhog Day guy: "Well, she might be okay"

    [seconds later, a 28-foot dive boat comes careening from up above, landing directly on her and the coral]

    Balding Groundhog Day guy: "… maybe not now"

  5. If you can explain something to me. Normally the further in the photo the more detail is lost to the camera sensor because of the water compared to clean still air above on land.
    Do I understand you start from a distance then cover that whole original scene again and again from different angles, and from all those raw gathered images, you compose one image 1-with the water removed, 2-All the detailed info combined and reduced in size/perspective to finally compose the original picture without water and the full detail further on as if it was one photo taken once, so something further is still small but detailed, like no quality was lost-?

    So what I am getting at, is you cannot just take one picture and work with it, simply because of the lost of quality -?

  6. I'm no expert, but I wonder if, via piezoelectrics, this technology could be applied to a diver's facemask so that his/her diving experience could be enhanced.

  7. I was speaking to a coworker about developing this kind of tech in order to REALLY see what lies within the depths. She looked at me like I was crazy. This video pops up so of COURSE I show my coworker and said, "…..Doesnt sound so crazy now does it?" She blankstared me and kept eating. 🤷🏽‍♂️ I'm happy. Lol

  8. Does anyone know of any good iPhone apps that can color correct underwater photos? I have a ton of pics from snorkeling in Cozumel on my phone that I'd love to do a color correction on to remove a bit of the blue and bring a more natural color to

  9. She should put some coral in a large swimming pool filled with water and take pictures. Then, drain the swimming pool and take more pictures. Compare. This would allow us to see if here algorithm actually makes it looks the way it should in dry air.

  10. How does this differ from color grading? We've had that algorithm since the 50's in analogue fashion and since the 90's in digital form.

  11. All good gifts and perfect gifts comes down from the Father of lights in who there is no variableness or shadow or turning Jesus said. And that faith comes by hearing and hearing by the Word of God. Fall in love with the Word of God ❤️📖🙂NKJV

  12. Just looks like she’s changing color values. I am disappointed. I thought it would actually remove the blue color entirely so you could see far and look like a desert.

  13. This is monumental holy shit— people don’t seem to realize just how unmapped and difficult the ocean is for us and this is a revolutionary step to understanding it better

  14. The formula is not complicated. There are underwater cameras with red light filters that remove the underwater effect. The filter is a matrix: so you just apply the matrix transformation to your live feed which is the algorithm

Leave a Reply

Your email address will not be published. Required fields are marked *