Comparison study of image segmentation techniques by curvature-driven flow of graphs

Abstract

In this paper, we deal with image segmentation by a curvature-driven flow of graphs. We focus on images, where the segmented objects can be represented by 1D graphs of functions. Compared to the usual direct approach, such a description benefits from certain advantages. Three methods to image segmentation are discussed and applied in terms of the graph formulation. Then, all methods are compared in the qualitative computational study.