Polar snakes: A fast and robust parametric active contour model

We present in this paper a way to perform a fast and robust image segmentation and to track a contour along a sequence of images. Our approach is based on a dynamic deformable model. More precisely, we revisit the physics based-model to show the benefit of using a polar description to model the contour, in particular to cope with the well-known initialization problem. Indeed, we show that this way to proceed leads to diagonal and constant matrices in the equations of the snake evolution yielding therefore to a faster algorithm. Experimental results on image segmentation and contour tracking validate the efficiency of this new formulation.