Abstract
Producing shoes which can fit the most people depends on an accurate foot sizing system. However, the current sizing standards in Taiwan (CNS 4800-S1093) can not fulfill this need. It is due to the fact that this standard is mainly adopted from the Japanese standard (JIS S-5037) and the latest edition of this standard was completed in 1998. There exists intrinsic foot shape differences between people in different countries as well as generations. In this study, we propose two 3D foot shape descriptors, i.e. the shape descriptor and the volume descriptor. The development of these two descriptors and the traditional foot anthropometry descriptor are based on a Taiwanese 3D foot shape scanning database, which contains 1835 male and 936 female right foot scanning data. Cluster analysis was applied to different gender groups to classify human foot shapes . Six different foot shapes were obtained. For each foot shape classification, we have developed a foot shape sizing system. A grading method was used to calculate the relevant shoe last measurements such as ball girth, arch height, and etc. Comparing to the current CNS sizing system, our sizing system can improve the coverage and reduce the size numbers. Thus, the manufacturer can apply these results for shoe last design with lower cost. This study also makes a comparison between different foot shape descriptors. The findings suggest that the volume descriptor, which contains 105 sizes with 97.1% coverage of male foot shape, has the best performance. This result can provide important reference for future research work.