Extended vector field convolution snake for highly non-convex shapes segmentation

Snakes, or active contours, are one of the major paradigms in image segmentation and they are extensively applied for the processing of the biomedical images. With the vector field convolution (VFC) as external force, they have a larger capture range and the ability to progress into concave boundaries. However, when we have to deal with highly non-convex shapes, the VFC field forms an area where the forces point in opposite directions and the snake stops before getting into the concavity if it is initialized outside. This field could be tailored by adding an anisotropic term that has as the result a displacement of the vector field upon a certain direction. We use this feature to drain the vector field out of the highly non-convex boundaries and we propose in this paper an image driven mechanism to generate this anisotropic term.