Compensation for distortion in the imaging process for 3-D surfaces

Wong, K.H., 2002. Compensation for distortion in the imaging process for 3-D surfaces. PhD, Nottingham Trent University.

Preview

Text 10290342.pdf - Published version Download (46MB) | Preview

Abstract

Digital models play a crucial role in today's numerous industries, ranging from rapid- prototyping in manufacturing to digital animation in the entertainment industry. In CNC manufacturing, digital models, obtained via reverse engineering of existing objects, help reduce the time to production. After the data has been captured, it can be refined or manipulated to create either a replica or a modified version of the original object. This reduces the cost and time in reproducing a blueprint and physical prototype of the object. There are many methods used to obtain 3-D images of an object, including laser triangulation (non-tactile) and touch-trigger probe (tactile). Different methods are appropriate for different purposes but ultimately each provides a set of points in 3-D space, representing the surface of the object of interest. These digital models, known as range images, often suffer from distortions, depending on the type of sensor used and also the physical attributes of the object. Before such an object can be physically reproduced, these distortions need to be compensated.

The size and nature of the distortions in the measurements depend on the type of sensor used. Significant distortions can be generated by a single-perspective active triangulation laser sensor (non-tactile), also known as a point sensor, in which a beam of laser light is projected towards the diffusely reflective surface of an object. The spot image produced on the surface of the object is sampled by a single position-sensitive photodetector in the sensor and the distance of this spot is computed using triangulation. This method of 3-D imaging is fast gaining ground in industries requiring reverse engineering, for it provides a rapid, cost-effective and non-destructive alternative to tactile imaging. However, when these optical sensors are used for scanning objects, significant distortions may occur.

A detailed investigation has been conducted into distortions obtained from a single perspective point sensor. Small distortions are encountered due to noise and simple algorithms have been developed for smoothing. More significant systematic distortions have been found close to high curvature regions of an object, especially where there are inclined or near-vertical faces. This systematic behaviour can be classified into two types. One type is caused by secondary reflections from other parts of the object while the other results from occlusion of the returning beam by the object. However, it has also been found that incidence of such distortions depends on the orientation of the sensor with respect to the geometry of the object, whereas it is independent of the scanning direction. Another contributory factor to distortions is the quality of the surface of the object.

Compensation algorithms have been developed to correlate multiple range images, taken with different orientations of the sensor, and yield a complete image with minimal distortions. Error regions are identified based on the disparities between the different range images. Edge detection algorithms have been developed to recognise inclined faces of an object based on its geometry adjacent to the error regions. The detected edges are then used to combine those range images with the smallest distortions within the error regions. The algorithms developed have been evaluated on both simple objects and more complex freeform objects.

Item Type:	Thesis
Creators:	Wong, K.H.
Date:	2002
ISBN:	9781369325911
Identifiers:	Number Type PQ10290342 Other
Rights:	This thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with the author and that no information derived from it may be published without the author’s prior written consent.
Divisions:	Schools > School of Science and Technology
Record created by:	Laura Ward
Date Added:	12 Jul 2021 14:45
Last Modified:	22 May 2024 15:48
URI:	https://irep.ntu.ac.uk/id/eprint/43414

Actions (login required)

Edit View

Views

Downloads