Ziplock Snakes

  • Authors:

  • Walter M. Neuenschwander;Pascal Fua;Lee Iverson;Gábor Székely;Olaf Kübler

  • Affiliations:

  • Communication Technology Laboratory, ETH Zurich, Swiss Federal Institute of Technology, CH-8092 Zurich, Switzerland;Computer Graphics Laboratory, EPF Lausanne, Swiss Federal Institute of Technology, CH-1015 Lausanne, Switzerland;Artificial Intelligence Center, SRI International, 333 Ravenswood Avenue, Menlo Park, CA 94025, USA;Communication Technology Laboratory, ETH Zurich, Swiss Federal Institute of Technology, CH-8092 Zurich, Switzerland;Communication Technology Laboratory, ETH Zurich, Swiss Federal Institute of Technology, CH-8092 Zurich, Switzerland

  • Venue:
  • International Journal of Computer Vision
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

We propose a snake-based approach that allows a user to specify onlythe distant end points of the curve he wishes to delineate withouthaving to supply an almost complete polygonal approximation. Thisgreatly simplifies the initialization process and yields excellentconvergence properties. This is achieved by using the imageinformation around the end points to provide boundary conditions andby introducing an optimization schedule that allows a snake to takeimage information into account first only near its extremities andthen, progressively, toward its center. In effect, the snakes areclamped onto the image contour in a manner reminiscent of a ziplockbeing closed.These snakes can be used to alleviate the oftenrepetitive task practitioners face when segmenting images byeliminating the need to sketch a feature of interest in its entirety,that is, to perform a painstaking, almost complete, manualsegmentation.