Iterative partitioning and labelling of point cloud data

Abstract

Over the past few years the acquisition of 3D point information representing the structure of real-world objects has become common practice in many areas. This acquisition process has traditionally been carried out using 3D scanning devices based on laser or structured light techniques. Professional grade 3D scanners are nowadays capable of producing highly accurate data at sampling rates of approximately a million points per second. Moreover the popularisation of algorithms and tools capable of generating relatively accurate virtual representations of real-world scenes from photographs without the need of expensive and specialised hardware has led to an increase in the amount and availability of 3D point cloud data. The management and processing of these huge volumes of scanned data is quickly becoming a problem.