Authors:
Sukhraj Singh
and
Subodh Kumar
Affiliation:
Indian Institute of Technology Delhi, India
Keyword(s):
Shape Descriptors, Human Body Models, Anatomy, Mesh Validation.
Abstract:
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Geometric Computing
;
Geometry and Modeling
;
Interactive Environments
;
Model Validation
;
Virtual Humans and Artificial Life
Abstract:
Advanced and personalized Human Body Models (HBM) are increasingly important in human centered industry design such as passive vehicular safety analysis, using finite element and other methods. Often accurate HBMs are painstakingly constructed for median human dimensions, and then modified and re-sized using personalization algorithms for various applications. Personalization algorithms rely on various anthropometric measurements, and sometimes manual intervention, to deform the median HBM. The quality of a personalized
model is often defined in terms of local properties such as aspect ratio of finite elements produced. In some cases it is inferred by visual comparison with some ground truth model or by measuring the anthropometric errors with respect to known values. We seek to define the quality of deformation in anatomically suitable geometric terms, which can be automatically computed. To this end, we compare the deformed anatomical surface meshes with that of the median mesh in
a shape descriptor space. Shape comparison and matching is a well studied area. The tools devised for the same are largely application dependent. We present pipeline and a metric for validating anatomical surface meshes. It is a problem that has not been extensively studied, even
though general shape comparison and matching techniques abound. Our metric incorporates global and part based shape signatures. The main contribution of our work is to explore techniques suitable for comparison of anatomical meshes by non technical experts. We formulate a pipeline that needs minimal user intervention.
(More)