摘要: We propose an active contour (a.k.a. snake) that takes the shape of ellipse. Its evolution is driven by surface terms made two contributions: integral data over inner ellipse, counterbalanced outer elliptical shell. iteratively adapt to maximize contrast between domains, which results in a snake seeks bright blobs. provide analytic expressions for gradient with respect its defining parameters, allows use efficient optimizers. An important contribution here parameterization ellipse we define such way all parameters have equal importance; this creates favorable landscape proceedings optimizer. validate our construct synthetic and illustrate on real as well.