There are as many ways to 3D scan an object as there are to print one (possibly more). For large-scale scanning, you want to look up photogrammetry, where you take dozens of photos of each thing (in this case, each person) that you want to make into a model and run it through a program. When I say dozens, I mean between 30 and 50 photos, from every angle possible, with all photos having the entire person in the frame and with the camera almost perfectly straight. The most accessible photogrammetry program I've found so far is Autodesk 123D Catch. It's free, but it's slow and imprecise.
If you have an old Microsoft Kinect, there are other options like Scenect and Skanect, though I never had very good luck with it myself, it's worth a try.
Both the photogrammetry and Kinect methods are more an art than a science, and they require a ton of practice in order to get anything useful. Also, for scanning people, you have to be sure they stand
perfectly still while you scan/photo them, which can sometimes be pretty much impossible.
There are other options, multi-camera rigs and purpose-built laser scanners, but those cost a
whole lot of money.