Nonlinear image labeling for multivalued segmentation

A nonlinear processing method developed for image labeling overcomes or minimizes the problems faced by common region-based algorithms. The method involves nonlinear integration of topological and statistical methods. It involves image scanning or growing processes, and pixel-labeling processes. The selection of the input threshold is unnecessary for a multivalue view of the labeling results obtained by the labeling mechanism. The method gives correct segmentation results for synthetic and real images.