PROJECT TITLE :

A Dynamic-Shape-Prior Guided Snake Model With Application in Visually Tracking Dense Cell Populations

ABSTRACT:

Here, we present the DSP snake model, which we believe will help improve the overall stability of the point-based snake model by using dynamic shape-priority (DSP). At its core is an efficient unification of a variety of high-level priors into one new force term for snakes. First, a global-topology regularity is presented to solve the intrinsic self-intersection problem for snakes. The uneven distribution of a snake's snaxels over the contour is also taken care of, resulting in a satisfactory parameterization. This method is different from other methods that use learning templates or enforce hard priors, which strongly respects the model's deformation flexibility while maintaining a reasonable global topology for the snake. This technique may effectively prevent snakes from self-crossing or automatically untie an already self-intersecting shape, as demonstrated in testing. This model is also integrated with existing forces and applied to the extremely difficult task of tracking dense living cell populations. The DSP G-snake model has improved tracking accuracy by up to 30% compared to traditional model-based techniques.. Real-world cell datasets with dense populations and substantial displacements were used to demonstrate that the suggested approach outperforms modern active-contour competitors and current cell tracking frameworks in terms of accuracy and speed.