Abstract：
A mannequin head was digitized using a reference scanner (Scan in a Box) to acquire the reference mesh. Subsequently it was scanned with a structured light scanner (Einscan Pro HD), a stereophotogrammetry scanner (RayFace100) and a laser scanner (Proface 3D Mid) to acquire test meshes.Resulting meshes were delineated in four horizontal areas and discrepancies calculated for the complete face and different facial partitions. One-way Anova and pairwise comparisons tests were used to compare trueness and precision between scanners across different areas. Significant differences were detected among scanners for complete face (F (3, 27) =776, P ⟨ 0.01)) and for delineated face areas (F (11, 99) =200.1, P ⟨ 0.01)). Einscan had significantly higher accuracy for the complete face (P⟨0.01) and significantly higher trueness for each facial partition compared to other scanners. RayFace had significantly higher trueness when scanning the middle part of face compared to other facial parts. Proface had significantly lower upper facial third trueness compared to other facial parts. All scanners had accuracy levels below the 2.00mm threshold. Facial scanning accuracy was influenced per scanner used. Scanning trueness per device was influenced by location of surface area. All scanners had accuracy levels within the acceptable accuracy threshold.
摘要：
使用参考扫描仪(在盒子中扫描)将人体模型头部数字化以获取参考网格。随后用结构光扫描仪（EinscanProHD）扫描，用于获取测试网格的立体摄影测量扫描仪(RayFace100)和激光扫描仪(Proface3DMid)。在四个水平区域中描绘了所得的网格，并为完整的面部和不同的面部分区计算了差异。单向Anova和成对比较测试用于比较不同区域的扫描仪之间的真实性和准确性。扫描仪在完整面部（F（3，27）=776，P﹤0.01）和划定的面部区域（F（11，99）=200.1，P﹤0.01）之间检测到显着差异。与其他扫描仪相比，Einscan对完整面部具有明显更高的准确性（P0.01），并且每个面部分区的真实性均显着更高。与其他面部部位相比，扫描面部中部时，RayFace的真实性明显更高。与其他面部部位相比，前脸的上面部第三真实度显着降低。所有扫描仪的精度水平均低于2.00mm阈值。面部扫描精度受到所用扫描仪的影响。每个装置的扫描正确率受表面积位置的影响。所有扫描仪的精度水平在可接受的精度阈值内。
