Use this URL to cite or link to this record in EThOS:
Title: Automated Calibration ofMulti-Sensor Optical ShapeMeasurement System
Author: Ogundana, Olatokunbo Omodele
Awarding Body: Loughborough University
Current Institution: Loughborough University
Date of Award: 2007
Availability of Full Text:
Access from EThOS:
A multi-sensor optical shape measurement systems (SMS) based on the fringe projection method and temporal phase unwrapping has recently been commercialised as a result of its easy implementation, computer control using a spatial light modulator, and fast full-field measurement. The main advantage of a multi-sensor SMS is the ability to make measurements for 3600 coverage without the requirement for mounting the measured component on translation and/or rotation stages. However, for greater acceptance in industry, issues relating to a user-friendly calibration of the multi-sensor SMS in an industrial environment for presentation of the measured data in a single coordinate system need to be addressed. The calibration of multi-sensor SMSs typically requires a calibration artefact, which consequently leads to significant user input for the processing of calibration data, in order to obtain the respective sensor's optimal imaging geometry parameters. The imaging geometry parameters provide a mapping from the acquired shape data to real world Cartesian coordinates. However, the process of obtaining optimal sensor imaging geometry parameters (which involves a nonlinear numerical optimization process known as bundle adjustment), requires labelling regions within each point cloud as belonging to known features of the calibration artefact. This thesis describes an automated calibration procedure which ensures that calibration data is processed through automated feature detection of the calibration artefact, artefact pose estimation, automated control point selection, and finally bundle adjustment itself. The process of calibration artefact selection is discussed, with the objective of developing a low cost artefact, with appropriate geometric and material properties such as unobstructed viewing by sensors, low coefficient of thermal expansion and non-specular surface finish. Automated detection of calibration artefact features is investigated, for enhancing the ease, speed, and accuracy of calibration. A novel 3-D Hough transform based on an optimised sparse 3-D matrix model is described, including methods developed for efficient peak detection in the Hough accumulator space. The calibration results of a two-camera and two-projector optical SMS based on multiple poses of the respective calibration artefacts developed, are discussed. A comparison of usage of the calibration artefacts is also made in order to assess their practicable use in an industrial environment. Based on acquired shape data of one of the artefacts, calibration accuracy of about one part in 5,000 was achieved. In applications for product inspection and quality assessment, the measured data needs to be presented in a form that provides for visualisation on a computer. A method for efficiently tessellating the measured point cloud using sensor pixel neighbourhood information is described. This method provides for the presentation of the measured point cloud data in industry accepted file formats. In addition, the measured data of a component may need to be compared against an idealised model of the component e.g. a computer-aided design (CAD) model. Methods for matching the measured data to a CAD model are therefore also discussed. The results of calibrating a multi-sensor SMS at an industrial site are presented. In spite of a less well controlled environment, a calibration accuracy of about one part in 1,600 was achieved, with the SMS subsequently used as a valuable tool for measuring out-of-plane displacement during a series of structural tests.
Supervisor: Not available Sponsor: Not available
Qualification Name: Not available Qualification Level: Doctoral
EThOS ID:  DOI: Not available