Fig. 1
From: Segmental HOG: new descriptor for glomerulus detection in kidney microscopy image
Candidate glomeruli and line segments. Segmental HOG (S-HOG) is based on the boundary of the objects of interest. If the boundary of a candidate glomerulus (Panel (a)) is to be located, boundary likeliness is computed at every point on m(=36) line segments placed uniformly in all m directions (Panel (b)). The boundary likeliness is computed at n points on each line segment. The n locations are depicted with dots in Panel (b)