Terrain: The surface of the moon and the rocky planets, Mars in particular, is of great interest to anyone trying to explore our solar system. The surface must be known in as much detail as possible, for missions to land safely, or for any robotic vehicle to traverse the surface. But until now, methods for analyzing images from an orbiting spacecraft, for example, have involved massive workloads and computing power, with limited results. A project by a former doctoral student at the Niels Bohr Institute at the University of Copenhagen, Iris Fernandes, has changed the situation. By studying the Stevns Klint limestone formation in Denmark, I have developed a method for interpreting shadows in images, so that the exact topography can be extracted. The method is faster and less labor intensive. The result is now posted! Planetary and space sciences 218.
Human space exploration involves high levels of safety – so accurate images of terrain are assertive
The topography of any surface will create shadow when sunlight hits it. We can clearly see the nuances in images of the moon, for example, but we don’t know the altitude of the terrain. So we see the topography changing, but not to what extent! The ability to see even very small features is essential to ensure a safe landing or movement in, say, a rover. Not to mention the safety of the astronauts.
If the rover can’t see detail, it might get stuck in sandy surfaces or crash into rocks – being able to see interesting geological formations to find rich geological environments for research purposes is also a big help.
The old limitations of terrain scanning have now been eliminated
When satellites orbit a planet, they can take pictures of the surface at reasonable quality. But to create an interpretation of the terrain that is accurate enough to land expensive equipment or perhaps astronauts, there is still a lot of customized information that needs to be processed.
The method of using shadows was there before, but it was computationally inefficient and still had to rely on assumptions. The new method uses a more direct and accurate calculation, does not rely on a whole set of parameters to be entered into the computer, and can also calculate uncertainties and accuracy.
This method is fast, accurate and does not rely on any assumptions. Previously,” says Iris Fernandez, “If you ask the question: How accurate is the topography assessment—there wasn’t a really satisfactory answer.
Now the exact terrain has been revealed, and we can even pinpoint the uncertainties. »
Scientific curiosity can take you to amazing places
“I was involved in a project where we wanted to use images from Stevens Klint to model patterns on the surface. I even presented this method at a conference in Los Angeles. But the nuances are a challenge, because the algorithm ‘sees’ nuances as geological features.
This created a bias in the model. We had to find ways to remove the nuances, in order to remove the bias.
I’ve always been interested in planets and have known that the surface of the Moon has been studied. There aren’t many annoying features on the Moon, so this was perfect for eliminating bias.
When we filtered the blinds, we could see what they were hiding, so to speak—the shapes of the roof, explains Iris Fernandez.
Solving existing images introduced a new problem – and a new approach
When work began on the Moon, the discrepancy between the different resolutions of the images and the topographic data turned out to be enormous. In other words, a new problem arose. “How can we combine different data sources with different precision?
This presented a huge mathematical problem – and that’s really what the study is about.
This is where the old research stopped. What we did differently from previous attempts to solve this problem was that we focused on math and reduced it to a tricky math equation. Basically, to see if this equation can solve the problem.
And he did,” Iris Fernandez smiles. “You could say that we, my supervisor, Professor Klaus Musegaard, have found the mathematical key for a door that has been locked for many years.
The path to follow
The focus is now on improving the method further. Wherever data on rock formation in the Solar System is available, such as the Moon, Mars, asteroids, or others, the method can be applied to extract precise topographical details.
The images used for this mission could be images of satellites or even of the rovers themselves, currently on Earth on Mars – or of any future mobile robot.
The goals of performing a correct topographical analysis can be different, it may be the safety of equipment, astronauts, or the search for geologically interesting sites.
In other words, there is a wide range of possible applications. “It’s kind of computer vision,” says Iris Fernandez, “when a robot has, for example, some form of machinery to interact with the environment, the method can help with navigation or ‘hand-eye coordination,’ because it’s less ‘heavy’ in computation.” Hence faster.
I’m just imagining now, but an interesting feature might be to assess the roundness of small rocks, in order to find the previous presence of water.
The method shows the data to us as humans in a way that we intuitively understand – like images of stones’ rotation, which are easy to interpret. »
“Evil thinker. Music scholar. Hipster-friendly communicator. Bacon geek. Amateur internet enthusiast. Introvert.”