Author(s)

M. Honkanen

Abstract

This paper presents image analysis algorithms to reconstruct a three-dimensional gas bubble surface from the time series of two orthogonal projections and to also recognize bubble outlines when the bubbles are partly overlapping in the image. Overlapping bubble images raise a large problem for three-dimensional, tomographic imaging of bubbly flows. Here, a novel approach to recognize the outlines of irregularly-shaped, overlapping bubbles based on the overlapping object recognition algorithm is presented. The main objective is to investigate the accuracy of a back-light imaging technique to measure bubble properties from a) a single projection or b) from two, orthogonal projections. The time series data of the bubble surface area, volume and orientation, location, velocity and trajectory direction is obtained. Temporal fluctuations in the measured bubble volume reveal the measurement accuracy in the case of a bubble swarm rising from a single orifice. Keywords: image analysis, three-dimensional imaging, high-speed imaging, bubble sizing, overlapping object recognition, optical tomography.

Keywords

image analysis, three-dimensional imaging, high-speed imaging, bubble sizing, overlapping object recognition, optical tomography