Where scientific calculation, AI and reverse engineering process satellite images
“Aut Caesar, aut nihil! Be Caesar or be nothing! This is a space in which one has to be big.” This is how Jean-Michel Morel, mathematician and expert in image processing at France’s National Center of Scientific Research (CNRS), sees Kayrros.
Jean-Michel has been involved with Kayrros since it was just an idea, working with founders Antoine Rostand and Jean-Michel Lasry to determine whether the mass processing of satellite images would even be feasible at the scale that they envisioned. Now, he works as a scientific advisor to Kayrros, driving the vision of the company’s processing systems and bringing any questions straight into the lab for answers.
Satellite image processing requires a combination of scientific calculation and major computing power to autonomously extract targeted insights. In the space between the click of a camera and the interpretation of the image, scientific algorithms explore the intricacies of the image down to each pixel, extracting the cues that a human can use to derive insight.
“Satellite imagery processing is one of the biggest scientific observation tools in the world. Clusters of satellites carry all sorts of cameras that see not only what is visible, but can also see many things that our eye cannot catch,” Jean-Michel explained.
Kayrros uses satellite imaging sensors to detect methane in the air, take measurements of metal structures on the ground or detect heat coming from industrial plants, to name just a few examples. But the challenge is knowing what to make of this data. This is where tailored but automatic processing systems come in, by examining more images than is humanly possible.
Jean-Michel Morel’s science lab is the digital darkroom of satellite imaging, the middle point between raw imagery and meaningful data where images are treated by mathematics and algorithms en masse. But here, the treatment is not lighting or contrast; it’s reverse engineering and programming.
“In the darkroom you improve the contrast, reveal more details, make it brighter and remove noise from the imagery. For us, it’s algorithms and mathematics. We can do much more than what we could do in a darkroom, because the fact that the images are digital means that we can reverse engineer them to really understand them, bit by bit. We can exploit everything in the image. It’s automatic, it’s accurate, and it’s on a massive scale,” Jean-Michel explained.
After earning a PhD in Mathematical Sciences from the Université de Paris 6, a passion for the fusion of mathematics and imagery drove Jean-Michel Morel’s career in France’s top scientific research institutions—notably at Université Paris-Dauphine and Ecole Normale Supérieure Paris-Saclay—at a crucial time in the development of the field of satellite imagery processing. In the lab, Jean-Michel and his team work on a dynamic array of projects dealing with image processing, from creating textural graphics, such as building imagery to look like wood, to image reverse-engineering for Agence France Presse and the French military, and determining whether such images have been tampered with.
For Jean-Michel, working with Kayrros provides a unique angle by introducing commercial questions—as well as the crucial element of reality.
“In fact, the main point here is that this is real,” Jean-Michel said with a laugh. “The fact that algorithms will be used massively in a real-world setting is extremely stimulating for us, and pushes our research to the limit. It’s a win/win situation for us because we do better research, and it’s good for Kayrros because we work on all aspects of image and video processing that arise commercially.”
Here, asking the right questions is key on both sides.
“If you look at the problem that Kayrros is trying to solve, you have so much to look at so you must be able to focus on the right questions. This is what a company like Kayrros is supposed to do: find those questions,” said Jean-Michel.
Satellite imagery is a critical component of Kayrros, providing the foundation for asset observation on a large scale at low cost. But nurturing partnerships with the leaders behind the world’s most innovative scientific institutions is what brings these foundational elements of the company to life. Kayrros leadership pushes the boundaries of the company’s potential by introducing the world’s most advanced research, straight from the lab. And in the words of Jean-Michel Morel, it’s a space in which one has to be big.