Development of an improved view registration method with application in optical velocimetry

Abstract

View registration (VR) refers to the process of determining the orientations and locations of the images captured by one or more cameras. It represents a fundamental procedure underpinning a wide spectrum of applications, ranging from optical sensing, object identification, and image processing. Existing VR procedures typically rely on a set of images captured on a fixed calibration target (referred to as FVM, fixed view method, hereafter). This work reports the development of a new view registration procedure, referred to as the variable view method (VVM). The VVM procedure relies on a set of images taken on a calibration target at various orientations and/or distances. As shown in this work, both numerically and experimentally, the VVM can significantly improve the VR accuracy compared to the FVM, by more than 40% in the tests conducted here. Such enhanced accuracy is expected to benefit a range of applications, and a specific example is demonstrated involving three-dimensional velocity measurement based on particle imaging velocimetry.