Scaling Constant Estimation for Texture Segmentation using Level Sets

Proposed work is aimed at finding a method to estimate a scaling parameter λ because of which minor object irregularities in the target texture image are ignored by the level set function during the process of curve evolution. Here the texture segmentation is achieved by embedding the statistical moment features in to the Level set frame work implemented as per Chan – Vese approach. The scaling parameter estimated here is used for emphasizing the variances of intensities of inside or outside regions of the evolving curve, which is estimated from the histograms of the extracted moment features. Reasonably correct values of λ are estimated and are substantiated by the results presented in the further sections. .