3D Face Scanner Comparison: Visual and Anthropometric Accuracy Analysis

Abstract

Compared to traditional manual measurement methods, 3D face scanning has opened more design possibilities for face protection product development, such as masks and goggles, by acquiring the three-dimensional shape and dimension. Various scanning technologies have been developed with different levels of price, expertise, and functional sophistication. In order to appropriately utilize 3D face scan data for a particular purpose, it is crucial to understand the characteristics of each scanner and validate their accuracy. The purpose of this study was to compare the visual and anthropometric accuracy of 3D face scanners and explore the characteristics and applicability of each scanner. Threedimensional face data of nine participants were acquired with three scanners: Artec Leo, Structure Sensor, and Bellus3D FaceApp. Before scanning, seven landmarks were marked, and five measurements were taken manually. The visual accuracy of each scanner was evaluated by experts through a survey with real-time comparison between scan and real face. It included assessment for distortion of textured and non-textured scans, as well as the visibility of landmarks. Measurements were obtained using Rhino 7 and Anthroscan ScanWorX software, and the anthropometric accuracy of each scanner was compared based on the manual measurements. The results of this study presented the visual and dimensional accuracy of the face mesh creation of three scanners, providing a comprehensive review of possible uses in consideration of the cost and usability of each scanner.

Quantitative Fit Assessment for Smart Gloves

“…Dimensional accuracy refers to the accuracy of the dimensions of the 3D mesh volume produced by the 3D scanners while visual/texture accuracy refers to the accuracy of the visual surface pattern, which can include landmarking, applied to the 3D mesh. Measurement tool validation is often conducted prior to fit assessment to quantify 3D scanner accuracy and differentiate scan error from fit conditions [30], [31].…”