Point clouds give users a lot of detailed information on objects and their whereabouts
Interview Jisk Attema, Maarten van Meersbergen & Oscar Martinez-Rubi
Point clouds give users a lot of detailed information on objects and their whereabouts
Interview Jisk Attema, Maarten van Meersbergen & Oscar Martinez-Rubi

Point clouds promise to make cities smarter

Point clouds play a key role in making a city smart. However it is still difficult to use that data, because there is so much of it. The eScience Research Engineers of the Netherlands eScience Center came up with important tools, tricks and solutions to use point cloud data in a project about the ancient road Via Appia.

Information with the push of a button

How many lampposts does Amsterdam have? How many trees are there in the Vondelpark? Where does garbage end up in the city? Of course, it is possible to retrieve this information by counting it all in person. But this is a complicated, expensive and time-consuming way. Wouldn’t it be wonderful if one could see this information with the push of one button? Automatic object recognition would be a great way to make a city smart.

Wouldn’t it be wonderful if one could see this information with the push of one button?

For these problems point clouds are promising. They give the user a lot of detailed information on objects and their whereabouts. Point clouds are used, for example, in the second national height map of the Netherlands AHN2 (Algemeen Hoogtebestand Nederland), which is used a lot by architects and city planners. This data set consists of an impressive 640 billion points and makes it possible to recognize trees, buildings and sidewalks.

eScience Research Engineers are working on web visualization of the 3D point cloud of the second national height map of the Netherlands AHN2

However it is very complicated to use point clouds. There are for example so many data points that it takes a long time to download only a part of a Dutch street. It is of crucial importance to make the data more accessible.

21st century archeology

The eScience Research Engineers have been working on that. They’ve recently collaborated with archaeologists in a project about the Via Appia. Point clouds were acquired of this beautiful ancient paved road in Italy. “Archaeologists wanted to be able to look at the Via Appia and its objects, for example graves, from behind their desk. So they wouldn’t have to go to Italy for their research or look at less detailed pictures”, says eScience Research Engineer Maarten van Meersbergen (eScience Center).

The Dutch company Fugro collected point clouds of the road by driving through it with a laser scanner on top of a car. Archaeologists also provided pictures of several objects along the road (graves and other monuments). “We’ve used these pictures to create point clouds of the monuments. We also needed to know where the exact location of each monument was. A human can do this manually, but it is time-consuming because there are so many of them. Therefore we created a tool, so the computer could put it in its correct place”, says engineer Jisk Attema (eScience Center).

Thanks to the new tool students can upload pictures of monuments that will automatically be put in the right place on or near the Via Appia

Attema and his colleagues use both the pictures made of the monuments and the point clouds of the road. They show the same kind of objects: rocks, trees. “A human can see immediately what the object is. But it is difficult to teach a computer this”, says Attema. “Sometimes it only sees a part of a building and it doesn’t know by itself whether it is a tree, rock or monument. It is also important to know the scale of what is depicted on the pictures. The archaeologists have put a measuring stick next to the monument when they took a picture. This makes it easier for a tool to automatically know how big something is. It is a bit like a puzzle. Our tool uses a lot of information to put objects in the correct position.”Every summer students do research on the Via Appia. Thanks to the new tool they can upload new pictures of monuments and it will automatically be put in the right place on or near the street.

Detailed point cloud of a monumental object along the Via Appia

Filtering data

All eScience Research Engineers worked closely together with the archaeologists. Van Meersbergen explained to one of them that it is possible to give all objects above the surface level a different colour. For example to ‘paint’a tree yellow. “It turned out they were not interested in the trees at all, because they were planted in the seventeenth century. Therefore they asked us if it was possible to remove them. This was no problem, but we only knew what was really important to them after talking it through”, says Van Meersbergen.

The user can navigate through the Via Appia and look at the different monuments

His team created the visualization of the point clouds. Thanks to their work the user can navigate through the Via Appia and look at the different monuments. One can even take a peek behind the monuments and zoom in. It looks and feels a bit like playing a videogame for scientists.

The eScience Research Engineers have done an impressive job in visualizing the point clouds. It was also a challenge to make sure the enormous amount of data could be processed. Because so many points are gathered it is very important to manage it in a smart way. Otherwise it will, for example, take a very long time to download point clouds.

Seen from above: The Via Appia point cloud

“When you’re looking at the Via Appia you want to see in detail what’s in front of you. But you don’t need to see all the details of the things that are further up the street. This distinction is not in the data itself, we have put it in there. We call this a pyramid data structure. You get in high resolution what is in front of you and less density of what is further away. Therefore you don’t need to download all the points”, says Oscar Martinez Rubi.

Preservation of beaches

The solutions, tools and tricks of the engineers will help to manage and use point clouds. Not only for the cases like the Via Appia, but also for other applications. For example the tools that Attema and his team created can help to develop object recognition, like identifying lampposts and trees. The pyramidal data structure will make it easier and a lot faster to download point clouds, for example of the second national height map of the Netherlands AHN2. And of course the visualization could also be used to navigate through other point clouds.

Our work is open source and we encourage others to use point clouds

“Our work is open source and we encourage others to use point clouds. There are so many possibilities”, says Martinez Rubi. “I’m from Catalonia, where there are a lot of beaches. They are now using point clouds to check the amount of sand. So they know when more sand is necessary to keep a beautiful beach.”

Text: Robert Visscher
Photography: Visualizations by the Netherlands eScience Center

Stay up to date, sign up for our newsletter