As such, Richard decided to create his own system. “I was seeing articles and forum posts describing electronic finders (devices that use a camera and computer to determine absolute telescope pointing),” he says. “So I decided to see if I could build something I’d be happy using and sharing.”
Scoping the sky
Richard has based his project – called PiFinder – around a Raspberry Pi 4 computer and a Raspberry Pi High Quality Camera, with the idea being to create a device that would take images of the night sky to determine where a telescope is pointing by analysing the pattern of stars.
“I wanted PiFinder to be easily added to any type of telescope, requiring no setup aside from a clear view of the sky to indicate where the telescope is pointing,” Richard says. “I wanted to combine this ability to know where the telescope is pointed with an extensive catalogue of thousands of object locations to help an observer find objects in the sky with simple guidance on which way to push the telescope.” He achieved all of these things and more.
By having the camera constantly taking images, the PiFinder can get to work. “So long as the on-board Inertial Measurement Unit indicates the telescope is static, the camera is snapping images and feeding them to the solving algorithm,” Richard explains. “Depending on sky conditions and which camera is being used, it can take anywhere from 0.25 to 1.5 seconds to capture an exposure which contains enough stars to solve.”
Stars in his eyes
After an image has been taken, it’s processed using code released by the European Space Agency called Tetra3. “It implements a very fast system for extracting stars from an image and using distances between multiple sets of four stats in the image to produce a hash,” Richard says. “This hash can be quickly matched against a pre-computed database of star patterns to find candidate positions. Each potential position is then checked using other stars in the same image to verify the correct solution.”
Extracting stars from the image on a Raspberry Pi 4B computer takes just 100 milliseconds, while solving the image takes 44 ms. “This is all done on a separate thread so the next image can be acquired while the previous image is being solved,” Richard says.
“In practice, this means that the PiFinder can usually achieve one capture/solve per second. If using the Raspberry Pi Global Shutter Camera, it can achieve more than two captures/solves per second due to its much larger pixel size and low-light sensitivity.”
By cycling through several astronomical catalogues and even typing in the ID of a specific celestial body, users can use PiFinder to locate objects in the night sky, receiving simple guidance on which way to push the telescope in order to view them. Richard says it’s working well, which is why he released it to the public after seven months’ work. “I’ll continue to focus on making it as simple to replicate and build as possible as I get more feedback from others,” he reveals.
