STOCHASTIC ASSESSMENT OF TERRESTRIAL LASER SCANNER

摘要: The registration of point clouds that are acquired from multiple viewpoints is a fundamental task for laser scanner data processing. To register the simultaneous adjacent scans, every scan has special targets attached onto object(s) like plain retro-reflective, black-white as well such targets, which have well-known geometrical 3D shapes (e.g. spheres, planes, cylinders, etc.). These or used landmarks and their centre coordinates modelled in local coordinate system. transform all scans to reference scan, least squares solution was fitting algorithm find parameters model transformation. Although, realistic stochastic modelling still difficult accomplish practice, both functional models need be carefully defined. In this paper mixed square calculate transformation parameters. Using model, left right considered observations. advantage approach giving same target different weights two scans. A assessment procedure been developed take into account range between scan. comparison current study results other commercial available software (Cyclone 5.1, Australis Panda) indicated reliability estimated improved.