The Internal Revenue Code has 3.8 million words — more than Shakespeare’s complete works. Trying to get that tax return filed while sorting through all the codes is a yearly frustrating task. Along the way, someone decided that more was better. Sometimes more is less.
Now we have LiDAR scanners collecting 1 million points per second and software with more than 250 proprietary file formats trying to keep up. Typical clouds are many billions of points. Except when you are working on a large project, this is far too many. Management of point data sets is already a concern for clients looking for catch-all applications, which rarely handle much of anything.
When it comes down to it, you can decimate (thin) the point cloud any way to match the end deliverable’s intended use. For bridges, that highway data in between structures in not needed. You are only looking for the area approaching and under the bridge. For highway pavement collection, these 360-degree scanners also capture overhead features such as power lines, tree canopy, etc., adding to the file size; so in the final deliverable you classify the point data to ground, low, medium and high vegetation. This classification allows the user to view the point data in various ways, such as only the ground plane, when working on a digital terrain model, or just the low vegetation when locating power poles or signs.
Depending on the application, point densities can range from one point per square meter to thousands of points per square meter. Users interested in bare-earth information (1- or 2-foot contours) typically can utilize a lower point density, whereas other applications, such as pavement management or feature extraction, require thousands of points per square meter. There are software algorithms for intelligent cloud reduction of redundant points, some of these algorithms can be found for free.
So why is it we are led to believe we need 1 million points per second to post-process in the name of efficiency when less will suffice? It is a tough question because, on one hand, you are already on site. Getting to the project site is one of the most costly tasks of a given project. The actual data collection is very efficient, so what’s a few extra billion points until it comes time to process those points and deal with them in a useful manner? On the other hand, if you collect too few points, it is a sad moment back at the office when you realize the data density does not support the intended use. So, in closing, understanding the intended use of the data is key in planning the data acquisition. That goes for accuracy requirements as well as point density.
To be more or not to be more? That is really the question.