What are point clouds?

Punktwolke point-cloud

WHAT ARE POINT CLOUDS?

Here’s an easy-to-understand introduction to the topic of point clouds. We answer the following questions:

Basics: What is a point cloud and how is it created?
What are point clouds used for?
How to work with point clouds? 

Punktwolke point-cloud

Basics: What is a point cloud and how is it created?

A point cloud can best be explained with the help of a gadget that reached the peak of its popularity in the 00s and is now primarily used for presentations: the laser pointer. The laser pointer can be used to illuminate a precise point in a straight line. If you know exactly where the laser pointer is located in the room, you can also exactly locate the point that is being illuminated. After all, the laser beam is straight as a die and thus makes it possible to calculate the exact position of the point in space in relation to the origin (the laser pointer). Geodetic points in surveying are also measured according to this basic principle, only it’s a little more complex. Instead of a laser pointer, special tachymeters are used for this purpose. 

And what does this have to do with point clouds? Quite simple. In addition to total stations, laser scanners have been used more and more frequently for surveying in recent years. These also work in the same way as our laser pointer, except that they can measure thousands or even millions of points simultaneously. Taken together, all the measured points constitue the point cloud. 

blank

What are point clouds used for?

Point clouds contain an incredible amount of information because every single point in the point cloud has its own X, Y, and Z coordinates. If we scan a staircase with a laser scanner, for example, we can use the resulting point cloud to determine exactly how straight the individual steps are, where the steps are worn and how high the deviation from the construction standard is. Accordingly, laser scanners and the point clouds they produce are always used when you want to document existing structures precisely, for example to digitise, measure or modify them. 

 

blank

 
For the design and optimization of production parts, for example in the automotive industry, very high-resolution hand-held scanners are usually used. They can capture even the smallest details and deviations. This allows a digital twin to be created on the PC. With the help of the twin, new prototypes can then be created and improvements digitally simulated and tested.

In the construction industry and as-built documentation of buildings, various scanners are used. Depending on the object, mostly terrestrial or mobile laser scanners, sometimes even drones. They are used to scan buildings for a variety of reasons, e.g. to plan an extension or renovation, to optimally position new equipment in production buildings, or to document the construction progress of various building projects. 

These are just a few examples of applications. Point clouds are used wherever objects need to be precisely captured and digitized. Depending on the area of application, different laser scanners are used. They can produce different accuracies and point cloud sizes. 

 

How to work with point clouds?

How to work with point clouds depends above all on what goal you are pursuing. As already mentioned, different laser scanners are used for different areas of application. The same applies to the software used to evaluate the point clouds. Our software solution, Origins (Pro), for example, is mainly used when existing buildings or landscape structures are to be digitally captured.

Regardless of which hardware and software solutions are used, there is an important step between the acquisition and the evaluation of the point cloud data: the registration.

 

Point Cloud Registration

During registration, individual scans or individual “sections” that were captured with the laser scanner are merged into a point cloud. If you want to register a complete building, for example, you often set up the laser scanner in the different rooms and scan them. Of course, in the end, you don’t only want to have individual scans of the different rooms. You prefer simply one large point cloud in which all the scans are available and linked together. To achieve this, you have to register the scans.

Behind the registration usually stands a rather complicated mathematical process. The accuracy of the data produced by the registration depends on how well the laser scanner captured the environment on site and how reliable the registration software used is. Fortunately, this process has become easier and easier in recent years. If you use a mobile laser scanner, for example, you often no longer have to make individual scans. You can simply walk through or around the object with the scanner. With this type of scanner, registration is also usually fully automatic and provided together with the hardware. The user does not need much know-how. The disadvantage here, however, is that mobile laser scanners are currently often not able to deliver as precise results as terrestrial laser scanners, i.e. scanners that are set up stationary and take individual scans one after the other.

 

blank

 

There are different methods to register a point cloud. The best known are cloud-to-cloud registration, target-based registration, or plane-to-plane registration. Which method to use depends on many different factors, e.g. the laser scanner used, the desired accuracy, or your own preferences. Especially for newcomers, it is, therefore, advisable to have the scanning and registration carried out by experts. They not only register the point cloud but also “clean” it in most cases during the registration process. This means, for example, that duplicate scans or “noisy” areas of the point cloud are removed or the point cloud is professionally “thinned out” to reduce the file size.

 

Importing, processing, and passing on point cloud data

If you receive registered point cloud data, there is usually one more stumbling block to overcome before you can take measurements and create digital 3D models from the data – importing the data. 

There is not just one file format for point clouds. In general, each laser scanner works with its own file format. Different software for processing are often using their own formats as well. As a result, there is hardly any software that can import all native file formats from the different laser scanners and processing software. We are very proud that our Origins (Pro) software can read and import over 25 different point cloud formats and export over 20 different formats (point cloud formats and others). However, even though we provide one of the greatest diversities on the market, it still doesn’t represent all native data formats. So what is the best way to deal with the different file formats?

Open exchange formats such as .las, .laz, .e57 or .xyz offer a solution to this problem. These file formats were developed by independent parties to solve the problem of data transfers. The .e57 format in particular has virtually become the industry standard. Almost all registration software of the laser scanner manufacturers can output the format and processing software for point clouds can also read the format. Therefore, in most cases, the surveyor will hand over the point cloud in .e57 format.

The disadvantage here is that the .e57 format, in contrast to the native formats, is less well compressed. Therefore it requires more storage capacity in comparison. However, the large amounts of data are normally no problem for point cloud software. After all, they were developed specifically for the processing and evaluation of point clouds. They can be used to carry out measurements and other evaluations. With Origins (Pro), for example, you can also create automatic floor plans that can be vectorized and much more

 

blank


However, if you want to create a 3D model from the point cloud, you need to employ a BIM or CAD software. These were not originally developed to handle point cloud data. Accordingly, many of these software still have major problems processing the data. Some CAD software, such as Autodesk Revit or AutoCAD, cannot read .e57 or other common point cloud formats. In order to use these software, the data must be converted again into the Autodesk formats .rcp or .rcs. Other CAD software cannot read point cloud data at all or can only import small amounts of data at once, which means that the point cloud has to be “split” again and imported in parts. These are all very tedious and time-consuming tasks.

To avoid this effort, the data is often first pre-processed in a point cloud software and then further processed in the CAD software. For example, Origins (Pro) can be used to create floor plans and vector lines, which can then be transferred to the CAD software in the correct position and with all the important 3D information in .dfx or .dwg format. These formats can be processed by almost any CAD software and require much less storage capacity than the entire point cloud. Of course, there are now also plugins for the most common CAD software that can transfer the 3D information from the point cloud software directly to the CAD software.

In conclusion, it can be seen that the acquisition and processing of point clouds, right up to the creation of a 3D model, still requires a great deal of expertise and know-how. Especially the amount of data and the data exchange between the different systems is still a challenge. Fortunately, a lot has already been done in recent years to simplify this process, also known as scan-to-BIM. We are also working every day to be able to import more data formats into our software and to simplify the handling so that even beginners can work with point clouds.

Do you have any further questions about point clouds or would you like to test our software yourself? Feel free to send us an e-mail to: support@pointcab-software.com. We would be happy to advise you in a personal meeting without obligation.