Joint segmenting and tracking densely packed cells using dynamic-GVF based snakes
Yao Lu, Sha Yu · 2018
This work develops a very generalised and efficient cell-tracking framework that is based on the combination of a new repulsive snake model and the dynamic gradient-vector-flow (DGVF) technique. To the best knowledge of the authors, the introduction of DGVF into dense-population based tracking is the first. And, by integrating the proposed repulsion mechanism into snakes, incorrectly merged segmentation can be more effectively prevented compared with existing schemes. Our experiments included very challenging datasets with densely packed cells and poor image quality. In spite of that, the proposed approach has demonstrated excellent performances with reported tracking accuracies in the range [85.23% ~ 93.15%].