From laser point cloud to surface : data reduction procedure test

Abstract

Data of object surface for reverse engineering can be captured either from contact or noncontact method. Laser scanning (non-contact method) usually gives high accuracy of captured data, however it also produces high point density. The difficulty of processing million of points captured by laser scanning device is the main problem to generate useful surface model. Data reduction process should be performed to reduce the number of points while maintaining the accuracy. This paper evaluated data reduction procedures on point cloud from laser scanner device. The procedures comprise of noise filtering, filter redundancy and point sampling method. Each procedure was tested using two types of sample models (freeform and primitive). The results show that the combination of filtering methods help to reduce the amount of original point cloud while maintaining the detail of surface