Handheld 3D Scanner: Based on Raspberry Pi
I’ve had some fun with photogrammetry recently and wanted to try out other techniques for 3D scanning. I’ve actually sort of accomplished S.L.A.M. during a hackathon using the PlayStation 4 camera and the Xbox One Kinect already, but I wanted something portable! With the new Raspberry Pi 4 debuting this year, I can finally accomplish this… sort of… Inside this 3D printed case is a Raspberry Pi 4 and a Intel RealSense D415 camera. The D415 is a depth sensing camera, using active stereo IR cameras and an internal processor to generate depth map data. It also has an ordinary RGB camera as well. That’s a lot of data to push through, hence why the newest Raspberry Pi 4 this year is important, because it has USB 3.0 ports that can handle all that data. (the ARM processor inside also has a faster architecture than before) How does the depth sensing work? Have you read my post on photogrammetry yet? There are some overlaps between the techniques. There are two IR cameras spaced apart, and it can see two very similar but shifted images. The processor finds interesting points (i.e. feature extraction) in the two images that are similar and match them up, the coordinates of these points can be used to calculate the 3D coordinates of those points relative to the cameras (i.e. binocular disparity). Why infrared? When there are not a lot of interesting features for the processor to look at, the…
