Shape analysis of paired landmark data

Abstract

This paper deals with the statistical analysis of paired landmark data using the exact distribution theory for the shape of planar correlated Gaussian configurations. Various properties of the paried shape distribution are investigated, and some special cases given. This work is a natural progression of what has been done on only ‘univariate’ shape analysis. However, one of the fundamental biological applications in shape analysis is in analysing shape change, and this work offers a possible solution. We illustrate our method by examining shape change in rat calvaria.