Simultaneous Tracking of Multiple Objects Using Fast Level Set-Like Algorithm

A topological flexibility of implicit active contours is of great benefit, since it allows simultaneous detection of several objects without any a priori knowledge about their number and shapes. However, in tracking applications it is often required to keep desired objects mutually separated as well as allow each object to evolve itself, i.e., different objects cannot be merged together, but each object can split into several regions that can be merged again later in time. The former can be achieved by applying topology-preserving constraints exploiting either various repelling forces or the simple point concept from digital geometry, which brings, however, an indispensable increase in the execution time and also prevent the latter. In this paper, we propose more efficient and more flexible topology-preserving constraint based on a region indication function, that can be easily integrated into a fast level set-like algorithm [Maska, Matula, Danek, Kozubek, LNCS 6455, 2010] in order to obtain a fast and robust algorithm for simultaneous tracking of multiple objects. The potential of the modified algorithm is demonstrated on both synthetic and real image data.