A Study on Approaches to Estimate Body Dimensions: Stature as an Example

Abstract

A large-scale anthropometric database [B. Das and A.K. Sengupta, 1996] was built in Taiwan in the late 1990s and was published in 2002. The procedures for collecting anthropometric data are usually complicated and costly in terms of resources such as workforce, time, and money. According to previous experiences and surveys among the designers and engineers, most practitioners do not know how the old anthropometric data may be converted into applicable new ones when updated data is unavailable [E.M. Wang et al., 1999]. Therefore, it is indeed a significant undertaking to develop methods that can easily convert old data into new ones easily especially with minimal errors. This study used statistical regression analysis and artificial neural networks (ANN) to estimate body dimensions and verified their effect. Subsequently, the estimation of stature built through stepwise regression was more accurate, convenient, and available compared to the other which was built through artificial neural networks.