We present a new fast active contour for images in 3D microscopy. We introduce a fully parametric design that relies on exponential B-spline bases and allows us to impose a sphere-like topology. The proposed 3D snake can approximate blob-like objects with good accuracy. The optimization process is remarkably fast. Our technique yields successful segmentation results even for a challenging data set where object contours are not well defined. This happens because our parametric approach allows us to favor prior shapes. This work comes with a companion software that allows extensive interactions between the end-user and our snakes through the intuitive manipulation of the few control points that fully characterize them.
Sahand Jamal Rahi, Vojislav Gligorovski, Marco Labagnara, Jun Ma, Xin Yang, Maxime Emmanuel Scheder, Yao Zhang, Bo Wang, Yixin Wang, Lin Han