Fig 3. Extracted features to link segmented objects.

Shown are two correctly segmented objects at time point t and a single segmented object due to an under-segmentation error at time point t + 1. To calculate cost terms, for each segmentation mask i at time point t the mask centroid p_i,t—shown as a cross –, a set of mask points ${\mathcal{Q}}_{i,t}$—shown in a lighter shade—and a bounding box ${\mathcal{B}}_{i,t}$—shown as a rectangle—are extracted. The Euclidean distance between the mask centroid p_j,t+1 and the propagated mask centroid ${\widehat{\mathbf{\text{p}}}}_{i,t+1}$ is large, which can result in wrong links. The minimal Euclidean distance between the propagated mask centroid ${\widehat{\mathbf{\text{p}}}}_{i,t+1}$ and the set of mask points ${\mathcal{Q}}_{j,t+1}$, in contrast, is small.