Abstract: We propose a design framework for the mass customization of custom-fit hearing aids and earphones with scanned models of the concha of 310 participants which are converted into compatible non-uniform rational B-spline surfaces generated through 795 data points to perform the requisite statistical analysis. Then, the quality of the reconstructed surfaces is determined through analyses of the shape deviations, Gaussian curvature, and zebra texture. Subsequently, the shape of the concha of the participants is grouped into 29 clusters with a modified algorithm. The results are then compared with conventional approach, and it is found that the modified algorithm has the ability to better classify the participants, but still uses a small number of clusters. Furthermore, the shape deviations between the samples and clusters are performed to validate the reliability of the clustering results and the importance of classifying the shape of the concha. Finally, a wear trial and simulation test are carried out to validate the wear comfort of a designed ear piece based on the average shape of each cluster. The experimental results show that the average shape obtained as per the cluster is capable of representing the common geometric properties of their corresponding members, and could thus be used as a reference in designing mass-customized ear-related products. The method in this study is superior to conventional methods that rely on sparse results for shape classification because it takes into account the intricate geometric shape of the concha.
 

Auricular concha; Surface reconstruction; Shape clustering; Mass customization; Custom-fit