Eliminating center discrepancies between simultaneously captured ILIDS and PIV images by means of direct homography estimation

Abstract

Abstract Interferometric Laser Imaging for Droplet Sizing (ILIDS a.k.a. MSI or IPI) requires the objective lens to be defocused so that fringe patterns can be imaged. When two cameras are used (e.g. to perform simultaneous PIV and ILIDS measurements or to assist in the detection of overlapping droplet images) this defocusing introduces a distortion that thwarts an accurate calibration of the two cameras and makes a successful registration of the two images impossible. We show that to overcome the obvious difficulties presented by empirical ad-hoc estimates of this “center discrepancy” distortion, existing feature-based registration and/or point set registration algorithms can be used on the images to find the correct homography directly. This approach eliminates the need for camera calibration and leads to greatly improved matching between images.