Many of the automated classification schemes that I have highlighted in the past use an iterative process to classify regions within the LiDAR data. In this paper, the authors present what appears to be an impressive alternative to the time consuming algorithms of other researchers.
This paper presents a classification approach of single return LIDAR data that uses an area growing technique to extract patches based on neighborhood height similarity. The extracted patches are classified according to its area into buildings, vegetation, and ground.
The authors also claim to provide the ability to easily tune the parameters to obtain better results. A short paper that deserves a look.