Space, the final frontier… In 1986, the SPOT satellite rocketed into orbit and turned its cameras toward Earth. Snapshot after snapshot, it built up a monumental stock of images. There was just one problem: among the millions of photos contained in the database, how could the image requested by a buyer be rapidly and accurately located? This was the unprecedented challenge tackled by Capgemini at the request of CNES.

Kourou, February 1986…

On February 22, 1986, the Ariane rocket was launched into the stratosphere from the space center in Kourou, French Guiana. Aboard was a 1,750-kilogram machine bearing the curious name of “SPOT-1”. It was the first European satellite for observing the Earth.

Three days later, the first color image appeared on the processing screen. The resolution was much better than the images taken by Landsat, the competing American satellite. The number of photographs quickly reached hundreds, then thousands, each one representing a tiny portion of our planet. It was a great technical success, but also the start of a new business venture.

Store, identify, process: a technical challenge in three words

The French national space agency (CNES), the SPOT satellite’s parent company based in Toulouse, decided to sell these images from space, to the military of course, but also to public institutions, farmers, meteorologists, and even major retailers planning new store locations. The agency created a special subsidiary dedicated to the activity called Spot Images. But it was still lacking one crucial tool: image processing software. That’s where Capgemini stepped in.

“The Group was a long-time partner of CNES. Our relations were excellent and built on trust. In addition, we had already processed images for the heavy equipment industry and were well-known in IT management, which is essential when setting up a marketing system. These were all elements that contributed to CNES choosing us,” explains Alexandre Haeffner, who then directed Cap Sesa Regions, the Capgemini entity that won the contract. CNES tasked the Group with creating software capable of storing, identifying, and processing image orders. It was a straightforward mission on paper at least. However, the Spot Images database was massive, containing several million images that needed to be sorted rapidly—an entirely new challenge at the time.

Farfetched solutions…

“You have to remember what the computer looked like at the time. The first Mac computers had just been released with black-and-white displays. Storing images was extremely costly and systems for consulting photo databases were still in development. Managing several million images was totally unheard of,” continues Alexandre Haeffner.

The Group’s 30 engineers assigned to the project had to invent everything from scratch, or just about, starting with the most essential aspect: precise classification criteria.

The task was far from easy. Everyone at Spot Images had their own ideas on how to proceed. “Since our task was unprecedented on this scale, we got some pretty outlandish suggestions on how to classify: by size, orientation, color, place… All these ideas had the same drawback: they did not allow for a precise image to be located,” states Haeffner.

In the end, Capgemini developed and installed a flexible solution combining several types of data. “It was the first time in Europe that anyone had developed an image recognition system based on several extremely precise criteria,” concludes Haeffner. The solution would emerge as the standard for this type of software, enabling Spot Images to become a world leader in marketing images from the sky.