Research On Personalized Dynamic Digital Human Model Generation Method Based On Scan Data

As an important tool in Computer-aided ergonomic, digital human models canbe applied in the areas of product and system design, ergonomic analysis,performance assessment and troubleshootings. Traditional feature-based digitalhuman modeling techniques require anthropometric statistical rules as externalconstraints, and usually a large body templates database is needed. Thedevelopment process is long and costly, in the meanwhile the fidelity of theconstructed model is not guaranteed. In recent years, with the development of3Dscan technology, body scan has been used in digital human modeling, these bodyscan based modeling methods don’t call external constraints or body templates, andthe constructed model has a high fidelity.The existing body scan based digital human modeling methods mostlygenerate static models or dynamic models with simple kinematic performance.Aimed at these problems, this paper has presented a personalized dynamic digitalhuman model generation method based on body scan. The method consists of modeldesign and model construction.Firstly, the design of the digital human model has been presented. The wholemodel consists of two layers: the interior skeleton model and the exterior surfacemodel. The multi-articulated skeleton model has many joints and degrees offreedom, this model is used to the generate the posture of the whole body. Thesurface model composing of triangular mesh is used for the body shaperepresentation. In the motion control part, the skeleton model performs rigid motion,and the surface model performs rigid and non-rigid motion.Secondly, the model construction method is introduced. The skeletallandmarks are identified from the scanned data to calculate the position of joints,and the skeleton model is generated. Then for each segment of the skeletal model,points are selected form the scan data by intersecting a set of radial lines with thetriangle faces of the body scan, forming the surface part of the segment.At last, based on the generation method presented in this paper, a digitalhuman modeling software has been developed, the software can generate the digitalhuman model from scanned data and realize its motion control and graphicvisualization. Using the software, several digital human generation experiments arecarried out and the results have been tested and analyzed.The experiment analysis shows that the digital human model generated withthe proposed method have the advantages as follows: the model has a good motionperformance, it can simulate complex human motions and performs well in skin deformation; the model has a high geometric similarity with the real object; themodel requires less amount of data and calculation and supports real-time motioncontrol. The model can be a useful tool in the research of motion simulation andposture analysis in computer-aided ergonomic.