8th International Conference on
Computational Harmonic Analysis

September 12-16, 2022

Ingolstadt, Germany

"Split-and-Merge Segmentation based on Graph Wedgelets"

Erb, Wolfgang

Graph wedgelets are novel adaptive building elements for the sparse decomposition of graph signals in geometrically meaningful, wedge-shaped, subregions. In this work, we study the usage of graph wedgelets as a splitting method in a combined split-and-merge scheme for the segmentation of images. We combine adaptive wedgelet splits of images with a simple and classical merging strategy for subregions and obtain in this way an efficient and robust segmentation of relevant domains in images.

« back