This researcher created an algorithm that removes the water from underwater images
Audio Brief
Show transcript
This episode covers Derya Akkaynak's groundbreaking Sea-thru algorithm for physically accurate underwater imaging.
There are three key takeaways. First, standard underwater photos are scientifically unreliable, but Sea-thru reveals true colors for accurate species identification and coral reef health assessment.
Second, the algorithm provides cleaner, accurate data, improving machine learning models for automated analysis of vast underwater datasets.
Third, obtaining reliable, repeatable visual data is crucial for quantifying climate change impacts, such as coral bleaching, on marine systems.
Ultimately, Sea-thru marks a significant leap for marine science and AI applications.
Episode Overview
- The episode introduces the primary challenge of underwater photography: images are often distorted, hazy, and have a strong blue or green color cast due to how water absorbs and scatters light.
- It showcases "Sea-thru," an innovative algorithm developed by oceanographer and engineer Derya Akkaynak, which digitally "removes the water" from images to reveal their true colors.
- The process for using the algorithm is detailed, involving a diver taking multiple photographs of a scene from different distances with a color chart for reference.
- The video highlights that this technology is a physically accurate correction, not just an aesthetic filter, with significant potential to advance marine science, particularly in studying coral reefs and the impacts of climate change.
Key Concepts
- Underwater Image Distortion: Water absorbs and scatters light, which removes color (especially reds and yellows) and creates a hazy effect known as backscatter. This makes it difficult to see the true appearance of underwater subjects.
- The "Sea-thru" Algorithm: A computer program that uses a physics-based model to reverse the optical distortions caused by water. It processes raw image data along with distance information to restore an image to what it would look like if photographed on land.
- Data Collection Process: To provide the algorithm with the necessary information, a diver must photograph the underwater scene from multiple distances. This, combined with a color chart placed in the scene, allows the program to calculate how colors degrade over distance and accurately reverse the effect.
- Physically Accurate Correction: The algorithm provides a scientifically valid restoration of color and clarity. This is different from photo editing software like Photoshop, which applies subjective, aesthetic adjustments rather than correcting for the physical properties of light in water.
Quotes
- At 00:07 - "I really see this as the start of the artificial intelligence boom in marine science." - Oceanographer Derya Akkaynak explains the groundbreaking potential of computational tools in her field.
- At 01:14 - "Because light, as it travels through the water, gets absorbed and scattered, colors fade away. That's why underwater images look so dull and distorted all the time." - Akkaynak describes the fundamental physics problem that her algorithm is designed to solve.
- At 02:55 - "This method is not Photoshopping an image... It's a physically accurate correction rather than a visually pleasing modification." - Akkaynak clarifies the key distinction between her scientific tool and standard image-editing software.
Takeaways
- Restoring true color in underwater images is critical for scientific research. This technology enables more accurate identification of species and better monitoring of the health of marine ecosystems like coral reefs, which is essential for studying climate change.
- Complex scientific problems can sometimes be solved with accessible tools. The "Sea-thru" algorithm works with photos from a standard consumer camera, making advanced marine imaging research more affordable and widely available.
- The underwater world is far more vibrant than we typically see. By removing the distorting effects of water, this tool reveals the brilliant, true colors of marine life, offering a more accurate and inspiring view of ocean biodiversity.
