Image segmentation with a Sobolev gradient method

Abstract The most effective methods for finding object boundaries in a digital image involve minimizing a functional over a set of curves or surfaces, where the functional includes internal energy terms for regularization and external energy terms that align the curves or surfaces with object boundaries. Current practice is to seek critical points of the energy functional by what amounts to a steepest descent iteration with the discretized L 2 gradient. Since the functional involves derivatives, a descent method with a discretized Sobolev gradient is likely to be much more efficient. We demonstrate this with test results for an implementation of a variational level set method for edge-based segmentation with active contours in two dimensions.

[1]  Chunming Li,et al.  Level set evolution without re-initialization: a new variational formulation , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[2]  John William Neuberger,et al.  Sobolev gradients and differential equations , 1997 .

[3]  S. Osher,et al.  Regular Article: A PDE-Based Fast Local Level Set Method , 1999 .

[4]  Anthony J. Yezzi,et al.  Sobolev Active Contours , 2005, VLSM.

[5]  Demetri Terzopoulos,et al.  Snakes: Active contour models , 2004, International Journal of Computer Vision.

[6]  W. Richardson Sobolev gradient preconditioning for image‐processing PDEs , 2006 .

[7]  J. Sethian,et al.  Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations , 1988 .

[8]  S. Osher,et al.  Regular Article: A PDE-Based Fast Local Level Set Method , 1999 .

[9]  Kishori M. Konwar,et al.  Fast Distance Preserving Level Set Evolution for Medical Image Segmentation , 2006, 2006 9th International Conference on Control, Automation, Robotics and Vision.

[10]  Guillermo Sapiro,et al.  Geodesic Active Contours , 1995, International Journal of Computer Vision.