The evaluation of scan-derived anthropometric measurements

Abstract

The use of 3-D whole body scanners has opened new opportunities for effective anthropometric data collection and its numerous applications. Meanwhile, there is a growing concern about whether the scan-derived measurements are comparable with the manual measurements. In this paper, an experiment was first conducted with 263 human subjects to evaluate the scan-derived measurements. By comparing the mean absolute difference (MAD) between the repeated scan-derived measurements against the interobserver error of the traditional anthropometric survey, the precision of the scan-derived measurements was found to be better than that of the manual measurements. As for accuracy, the results of the paired T-test indicated that some of the scan-derived measurements were significantly different from the corresponding manual measurements. In addition, most of the MADs between the scan-derived measurement and the manual measurement exceed the ISO 20685 criteria. The unacceptable accuracy was possibly due to the variation caused by the human subject. Thus, a mannequin was utilized for a second-stage evaluation of the scan-derived measurements in which the variation of human subject was eliminated. In contrast to scanning human subjects, the experimental results showed that the scan-derived measurements produced acceptable and satisfactory accuracy and precision while scanning a mannequin. As a result, the variation caused by human subject seems to be a critical influencing factor and needs to be minimized in 3-D scanning. Moreover, some important countermeasures are proposed for enhancing the performance of 3-D scanning, including designing supporting devices to reduce body sway, training the staff with adequate measurement skill, and increasing the scanning speed and resolution of the 3-D body scanner. © 2006 IEEE.