From Pixels to Voxels: Tracking Volume Elements in Sequences of 3-d Digital Images

This paper presents the extension of certain, fundamental digital photogrammetric concepts, theories and strategies to three dimensions, and their performance in an interdisciplinary application for 3-D image sequence analysis in a technical chemistry experiment. Specifically, we present an image analysis technique for examining the fine scale (Kolmogorov scale) variations of the mixing process in a turbulent flow. The objective is to trace the interaction of two flows by identifying their motions in a sequence of 3-D images obtained with a system based on a high-speed solid state camera. After briefly describing the imaging process and the particularities related to the capture of quasi-continuous 3-D image sequences, we focus on theoretical and implementational issues associated with feature tracking in 3-D image sequences. We present the extension of least squares matching from pixels, associated with 2-D images, to voxels, associated with 3-D images. The use of additional constraints of radiometric and/or geometric nature strengthens the matching solution. In addition, the large amount of data associated with 3-D image sequences in general, and the high and multidirectional velocities involved in this application in particular, make the devision of an efficient matching strategy quite important.