Abstract
This study established a 3-D craniofacial anthropometric database and applied it to the design of full-face respiratory masks and safety helmets. A photo-grating measuring system, which scans more than 100,000 3-D surface points in 94 seconds, was used to measure craniofaces. The database consisted of 1213 subjects, randomly selected from the Taiwanese worker population, but adjusted for gender and age group distribution. This database was then used to extract 39 1-D measurements for general-purpose design reference and to design full-face respiratory masks and safety helmets in 3-D datum format.These 39 1-D measurements, 31 point-to-point distances and eight arc lengths, were calculated based on a set of 24 anatomic landmarks. Among these 24 landmarks, 16 with apparent features were identified automatically and eight with fine details were identified manually.For the design of full-face respiratory masks, three typical 3-D head models were selected, which represent the norms of the small, medium and large size groups. These three sizes intended to fit best to the whole population. These 3-D head models were processed by sizing and selection phases. The sizing phase was to divide the whole sample set into three size groups by optimizing the dispersion of the face length and face breadth against allowable design tolerance. The selection phase was to select a “typical model” within each size group using form-difference analysis algorism. The typical model was the one with the smallest form-difference value within the group.For the design of safety helmets, sixteen 3-D synthesized cranial models were formulated in accordance with ISO 3873-1977 standard specification. The cranial model was the head volume above the head circumference at glabella level. The head circumference was used as the key measure for sizing, with a size interval of 10mm in range between 490mm and 650mm. The size group with the largest number of sample coverage was choosing as the based model. The base model was synthesized by superimposing all cranial surfaces on the center of head circumference. From the center, a series of radial lines each at 5o interval both in polar angle and azimuth angle. Each radial line intersects with all (superimposed) cranial surfaces, for each cranial surface there is a radial length. By weight-averaging all these radial lengths with a cosine weighting function (K=50(1+|cosθ|)), we obtained a series of points. Then by constructing a smooth 3-D surface based on these points, a synthesized cranial model was obtained. Finally, each of these other fifteen models was then graded by means of proportional ratio based on its maximal head circumference within the group.