Depth from Defocus Technique for High Number Densities and Non-spherical Particles

Abstract

The Depth from Defocus (DFD) imaging technique for measuring the size and number concentration of particles in a dispersed two-phase flow has up to now been restricted to relatively sparse particle densities and to identifying only spherical particles. The present study examines two advancements to the technique, widening its range of application significantly. The first advancement introduces an image processing procedure which can identify and size particle images which are overlapping. This increases the tolerable number concentration of particles which can be identified and processed within the measurement volume. The second advancement explores the possibility of determining the size and position of non-spherical particles within an observation volume. Both advancements build on recent theoretical work, utilizing not only the gray level of the blurred, out-of-focus images, but also the gradient of the gray level normal to the nominal particle or particle ensemble contour. This gray-level gradient is used to estimate the width of the blur kernel, which is assumed to remain Gaussian. These enhancements are experimentally validated using a dedicated apparatus in which particles of known size, shape and degree of overlapping images can be systematically varied. This experimental setup provides benchmark data to quantify the accuracy and limitations of the processing algorithms.