Image Segmentation by Size-Dependent Single Linkage Clustering of a Watershed Basin Graph
This work addresses the problem of efficient segmentation for large-scale 3D brain images, though it appears incremental as it modifies existing clustering techniques.
The paper tackles hierarchical image segmentation by introducing a size-dependent single linkage clustering method applied to a watershed basin graph, achieving quasilinear runtime suitable for large images, as demonstrated on 3D electron microscopic brain images.
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.