Stability and convergence of the level set method in computer vision

Chaudhury, Kunal N and Ramakrishnan, KR (2007) Stability and convergence of the level set method in computer vision. In: Pattern Recognition Letters, 28 (7). pp. 884-893.

PDF Stability_and_convergence_of_the_level_set_method.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Several computer vision problems, like segmentation, tracking and shape modeling, are increasingly being solved using level set methodologies. But the critical issues of stability and convergence have always been neglected in most of the level set implementations. This often leads to either complete breakdown or premature/delayed termination of the curve evolution process, resulting in unsatisfactory results. We present a generic convergence criterion and also a means of determining the optimal time-step involved in the numerical solution of the level set equation. The significant improvement in the performance of level set algorithms, as a result of the proposed changes, is demonstrated using object tracking and shape-contour extraction results.

Item Type:	Journal Article
Publication:	Pattern Recognition Letters
Publisher:	Elsevier
Additional Information:	Copyright of this article belongs to the Elsevier.
Keywords:	Level-set method;Tracking;Curve evolution;Difference equation;Numerical stability;Courant–Friedrichs–Levy condition
Department/Centre:	Division of Electrical Sciences > Electrical Engineering
Date Deposited:	08 May 2007
Last Modified:	19 Sep 2010 04:37
URI:	http://eprints.iisc.ac.in/id/eprint/10868

Actions (login required)

View Item