A nonlinear variational method for signal segmentation and reconstruction using level set algorithm

A nonlinear functional is considered in this letter for segmentation and noise removal of piecewise continuous signals containing binary information contaminated with Gaussian noise. A discontinuity is defined as points in time scale that separates two signal segments with different amplitude spectra. Segmentation and noise removal of a piecewise continuous signal are obtained by deriving equations minimising the nonlinear functional. An algorithm based on the level set method is employed to implement the solutions minimising the functional. The proposed method is robust in noisy signals and can avoid local minima. r 2006 Elsevier B.V. All rights reserved.