How can we improve the laser scanning of shiny surfaces which can give unwanted reflections ?

HUYNH, Cao-long

4. term

Vision, Graphics and Interactive Systems, Master

2010

2010-06-03

64

3D-scannere, der bruger triangulation, har ofte svært ved at måle skinnende eller semitransparente overflader som glas. Uønskede refleksioner kan få scanneren til at registrere falske punkter, som bliver til fejl i 3D-punktskyen (samlingen af alle målte punkter). Dette arbejde gennemgår flere løsninger og implementerer én tilgang baseret på lokal punktsky-tæthed (antallet af naboer tæt på hvert målepunkt). Punkter, der er usædvanligt isolerede, er sandsynligvis forårsaget af refleksioner og kan markeres og fjernes. For at sætte metoden i kontekst beskriver vi de grundlæggende principper for en 3D-laserscanner. Test af implementeringen viser, at teknikken effektivt opdager og fjerner falske punkter, men behandlingshastigheden kan være en begrænsning i nogle tilfælde. Yderligere optimering er nødvendig for at vurdere, om metoden er egnet til industriel brug.

3D scanners that use triangulation often struggle to measure shiny or semi-transparent surfaces, such as glass. Stray reflections can cause the scanner to record false points, which appear as errors in the 3D point cloud (the set of all measured points). This work reviews several solutions and implements one approach based on local point cloud density (how many neighboring points lie close to each measured point). Points that are unusually isolated are likely caused by reflections and can be flagged and removed. To put the method in context, we outline the basic principles of a 3D laser scanner. Tests of the implementation show the technique is effective at detecting and removing false points, but its processing speed may be a limitation in some situations. Further optimization is needed to assess whether the method is suitable for industrial use.

[This abstract was generated with the help of AI]

Download
View record in AAU Student Projects