This is one of many, many ways to generate a mesh from a point cloud using Meshlab.
The best possible technique if you're using a Kinect would be to start by using the image space pixel positions to build a regular triangulated mesh. Then from there you could process it and get rid of long triangles or clusters of noisy data.
If there are any advanced Meshlab users with tips on better meshing techniques, please leave a comment or contact me!
Loading more stuff…
Hmm…it looks like things are taking a while to load. Try again?