I am working for the City of Jacksonville and have been assigned to create an updated building footprints layer. These are just polygons with an elevation attached to them. With the sheer number of buildings that need to be updated, manually digitizing is not an option.
I am looking at two different ways of doing this, but I have only just finished my undergrad with a certification in GIS so there is only so much I know how to do.
The first involves lidar. I took lidar points and converted them to a 5x5 raster. I then subtracted the bare earth DEM from USGS from the "first pass" DEM raster created from the lidar points. I then use a con statement to basically select out bare ground and roads. My thought here is that if I select out everything that is above about 10ft (an average one story building) I would get raster coverage of all the buildings and could them change them to integers and convert them to polygons etc. But I run into trouble with the con statement as it selects out parts of buildings and I can't seem to get rid of all the ground data.
The second idea I have is to just use an aerial image and do a classification. The only thing is I do not know whether to use a supervise or unsupervised and how to go about it that way.
If anyone has suggestions or ideas, they would be greatly appreciated.