Transferring Landmarks to Individual Foot Using Deformable Template Models

Abstract

An efficient and novel method is presented to automatically mark foot shape anatomical landmarks in human body. 3D digital foot shape including anatomical landmarks is taken as deformation template. PCA and ICP algorithms are applied to make sample foot shape and template foot shape align and registration in the least-squares sense. Then initial corresponding point set between sample foot shape and template foot shape is found. Sample point set is obtained in the template foot shape by using method that combining uniform sampling and sampling based on distribution of curvature. Points in sample point set corresponding to template foot shape are taken as position constrained condition. Discrete Laplacian mesh deformation is used to make standard foot shape fit sample foot shape. Template foot shape is approached to sample foot shape in error range via iteration. Therefor, the mapping relation between template foot shape points and the corresponding points on sample foot shape is built. So the corresponding points of anatomical landmarks on template foot shape, on the sample foot shape, can be treated as anatomical landmarks on sample foot shape.