Alternative Methods for Estimating Plane Parameters Based on a Point Cloud

Abstract

Non-contact measurement techniques carried out using triangulation optical sensors are increasingly popular in measurements with the use of industrial robots directly on production lines. The result of such measurements is often a cloud of measurement points that is characterized by considerable measuring noise, presence of a number of points that differ from the reference model, and excessive errors that must be eliminated from the analysis. To obtain vector information points contained in the cloud that describe reference models, the data obtained during a measurement should be subjected to appropriate processing operations. The present paperwork presents an analysis of suitability of methods known as RANdom Sample Consensus (RANSAC), Monte Carlo Method (MCM), and Particle Swarm Optimization (PSO) for the extraction of the reference model. The effectiveness of the tested methods is illustrated by examples of measurement of the height of an object and the angle of a plane, which were made on the basis of experiments carried out at workshop conditions.

A Novel Type of Boundary Extraction Method and Its Statistical Improvement for Unorganized Point Clouds Based on Concurrent Delaunay Triangular Meshes

“…In the past two decades, three-dimensional (3D) laser scanning technology and its equipments have rapid developments and broad applications in many important fields, such as the reverse engineering [ 1 ], non-contact measurements [ 2 ], computer vision [ 3 ], and building reconstruction [ 4 ]. Point cloud dataset collected by 3D laser scanners is one of the most basic and popular data types in engineering practices [ 5 ].…”