The first step is to drastically thin the data using a PointCloudThinner, since we are only interested in a creating a boundary.

The second step is to use a PointCloudCoercer to conver the thinned point cloud to a multi-point.

The third step is to use a pair of Bufferers to create the boundary. The first buffers the aggregate multi-point by a distance larger than the max distance between the thinned points, creating a bounday with no holes. The second buffers the resulting polygon by the negative of the first distance to shrink the outer boundary back to the edge of the points.

This is a fairly efficient operation,and can process a 500 MB Lidar dataset in less than 30 seconds.


See the attached workspace (LidarConcaveHull.fmw) for details