Shape clustering as a type of procrustes analysis

Abstract

The ordinary Procrustes sum of squares is one of the most important measures in Procrustes analysis of shape. In this paper, we incorporate a competitive learning scheme into Procrustes analysis. We introduce a measure of distance between the landmarks of shapes for a competitive learning scheme. Thus, we present novel shape clustering as a type of Procrustes analysis. Using datasets of line drawings and outlines, we show that shape clustering performs well for shape classification compared with shape clustering founded on typical vector-based distances.