Three Dimensional (3D) Head Data Classification Based on a Local Shape Feature Description
Xiaomei Zheng, Jianwei Niu, Shixian Ding, Qinqin Zhou · 2012
Three Dimensional (3D) anthropometric data can be obtained more easily than decades before, but to process 3D data is most challenging and 3D shape classification based on local feature is one of the challenges.In this paper, a local shape description based on spin-image, combined with clustering method for 3D head data analysis, is proposed.Numerous methods have been proposed to extract local features of landmarks on 3D head data.But the method with high efficiency, good robustness and strong adaptability is not found out yet.An approach using spin image was adopted in this study to describe the 3D head landmark features.For each head, eight landmarks were selected and their responding spin-images were calculated.In this way, a head can be transformed from point cloud to a vector formed by spin images.K-means clustering algorithm was performed on the transformed data to classify the samples into seven groups.Totally one hundred 3D head data of young male Chinese have been analyzed to illustrate the applicability of the method.Classifying a population of 3D data using this method may provide a promising way to improve product design for fitting comfort.Whether the method in this paper is suitable for 3D data classification of other body segment deserves much more investigation