Classification of the road infrastructure objects based on mobile two-dimensional laser scanning

Abstract

The paper deals with classification of road infrastructure objects, particularly road surface and objects situated in surrounding environment such as facades of buildings and pole-like objects. The attention is paid to processing of three-dimensional models generated and reconstructed on the base of two-dimensional mobile laser scanning. There are different methods available to classify scanned objects; however, various shortcomings and limitations are usually identified in relation to their practical usage. Most classification methods applied to the three-dimensional models are based on range data processing. We present our three-stage classification algorithm that has been proposed and implemented with motivation to explore how range and reflectivity data could be used separately or in fusion to classify certain objects. The paper shows results of practical experiments based on real data recordings and proving usability of the proposed algorithm.