Abstract
Tracking algorithms used in conjunction with fluorescent time-lapse microscopy data typically assume a continuous signal where (a) background and keypoints are permanently visible and (b) independently moving objects of interest are permanently visible when not occluded. These conditions allow for registration and identity management algorithms to track independently-moving objects of interest over time. In contrast to such conditions, we consider the case of (a) transient fluorescent events and (b) objects embedded in a possibly imperceptible, almost rigid structure, which acts to constrain independent object movement. In a biological context, such events could for example represent cell divisions in a growing tissue. Such conditions motivate the merging of registration and tracking tasks into a fuzzy registration algorithm to solve the identity management problem. We describe the design and application of such an algorithm, illustrated in the domain of plant biology and made available as an open source software implementation.
