do you have anyway to distinguish the buildings? Such as a building footprint?

I don't have a conclusive answer, but if all you have is a LIDAR point cloud and need to extract buildings out of it there's a few things you can try. Keeping in mind that both of these will find "bumps"... anything that stands out from the terrain, so buildings, trees, parked cars, garbage bins, anything.

• Create contours and build polygons out of those, grouping by elevation value, dissolve the resulting polygons and the results should be your bumps.
• Build a TIN out of the point cloud and look for closed rings of triangles with a high slope value.

Both methods are time consuming and will most likely yield results that need to be interpreted further. I'm trying to find trees in a LIDAR dataset and would be looking at area and circularity as criteria.