Image Segmentation by Size-Dependent Single Linkage Clustering of a Watershed Basin Graph

TL;DR

This paper introduces a hierarchical image segmentation method that combines watershed basins and size-dependent single linkage clustering, enabling efficient segmentation of large 3D images.

Contribution

It proposes a novel size-dependent clustering approach on watershed basins with quasilinear runtime for large-scale image segmentation.

Findings

Effective segmentation of 3D electron microscopic brain images

Quasilinear runtime suitable for large images

Improved accuracy over traditional watershed methods

Abstract

We present a method for hierarchical image segmentation that defines a disaffinity graph on the image, over-segments it into watershed basins, defines a new graph on the basins, and then merges basins with a modified, size-dependent version of single linkage clustering. The quasilinear runtime of the method makes it suitable for segmenting large images. We illustrate the method on the challenging problem of segmenting 3D electron microscopic brain images.