Massiv mængde 3D data behandles på standard computer

Forskere på Max Planck Institut vil på Cebit demonstrere banebrydende software, der gør det muligt at behandle ekstremt store mængder af 3D data på en standard pc (in english).

3D scanners are used today in industrial quality control and for construction and urban planning.  These tasks create gigantic amounts of data that push even high-capacity computers to their limits, especially if the data is processed in real time. 

Researchers at the Max Planck Institute for Computer Science in Saarbrücken have developed a method of processing huge three-dimensional scans on a standard PC.  The researchers will present this new technology at the CeBIT 2011 in Hanover from March 1st through the 5th at the Saarland research booth (hall 9, booth B43).

- The 3D scanner scans the surface of an object and measures the position of every single point on its surface, and in doing so the scan of a building, for example, quickly creates several gigabytes of data, says Art Tevs, a researcher at the Max Planck Institute for Computer Science. 

If this method is used to acquire entire cities or large industrial facilities, the amount of data increases many times over. However, three-dimensional scans of objects can be used in a variety of ways, be it in architecture, urban development or industrial quality control.  Thus, the researchers in Saarbrücken were looking for new methods to process these enormous floods of data in real time. 

- With especially elaborate algorithms, it is now possible to process and examine extremely voluminous data from one scan, and this data can be interactively visualized and modified with a standard PC, Art Tevs explains.

The researchers in Saarbrücken are able to automatically identify recurring structures in objects, for instance, parking garages or medieval buildings, and reduce them to their basic component parts.

- The amount of data is then significantly reduced and can be processed more quickly, Tevs says to name one advantage.  Fragmentary data, which occurs if individual objects are occluded at the time an image is made, is also a challenge. 

- Our software can complete such deficient data.  If, for example, the wall of a building is scanned, and the window frames cannot all be clearly detected, we superimpose several partial pictures of frames in order to significantly improve the quality of the image, Art Tevs says.

The new software can also detect patterns in the objects.  The researchers use this information to virtually, with slight modifications, rebuild them. 

- We can analyze the structure and details of the component parts and then derive from them a virtual building plan of the object, the researcher explains.  Art Tevs is also a member of the Cluster of Excellence at Saarland University.  The computer scientists in Saarbrücken demonstrate how three-dimensional images of scanned buildings can be altered and enhanced.