Two-dimensional images of synthetic industrial diamond particles were obtained
using a camera, framegrabber and PC-based image analysis software. Various
methods for shape quantification were applied, including two-dimensional shape
factors, Fourier series expansion of radius as a function of angle, boundary fractal
analysis, polygonal harmonics, and corner counting methods. The shape parameter
found to be the most relevant was axis ratio, defined as the ratio of the minor axis to
the major axis of the ellipse with the same second moments of area as the particle.
Axis ratio was used in an analysis of the sorting of synthetic diamonds on a vibrating
table. A model was derived based on the probability that a particle of a given axis
ratio would travel to a certain bin. The model described the sorting of bulk material
accurately but it was found not to be applicable if the shape mix of the feed material
changed dramatically. This was attributed to the fact that the particle-particle
interference was not taken into account.
An expert system and a neural network were designed in an attempt to classify
particles by a combination of four shape parameters. These systems gave good results
when discriminating between particles from bin I and bin 9 but not for neighbouring
bins or for more than two classes.
The table sorting process was discussed in light of the findings and it was
demonstrated that the shape distributions of sorted diamond fractions can be quantified in a useful and meaningful way.
A Doctoral Thesis. Submitted in partial fulfilment of the requirements for the award of Doctor of Philosophy of Loughborough University.