High Precision Parametric Modeling Of Human Calves Shape Based On NURBS

With the rapid development of non-contact measurement technology such as 3D scanning technology,point cloud representation of 3D shape has been accepted as a popular description of surface,which gives birth to relevant digitization or modeling methods.However,common methods such as surface fitting or registration face great challenges in providing custom designed clothing products,especially the strong demand for high-precision modeling in medical and rehabilitation applications.Concretely,those methods large number of feature points to improve the smoothness in shape representation.This will inevitably lead to the balance between accuracy and computational efficiency.In addition,when it comes to fitting,the selection of feature points should conform to anatomical facts,rather than equidistant picking by algorithm.In order to overcome the above problems,this paper proposes a new parametric model of human calves based on the non-uniform rational B-splines（NURBS）using identified feature points based on human anatomy.Thanks to NURBS,this model is with merits of high-accuracy,good smooth in nature as well as higher efficiency.Details of this work are as follows:First,after data initialization and resampling,circumferential profiles of human calves were analyzed to obtain the feature cross-sections according to human anatomy characteristics.Curvature analysis was conducted then on contour curve of each feature cross-section,giving feature points that were then arranged to a rectangular mesh of data points for further NURBS fitting.Second,control points were then be calculated,based on which and along with knot vectors,the NURBS surface of human calves can be constructed.To test the validity of this parametric modeling,an error index was proposed based on a self-designed point-to-NURBS surface function,which is of high efficiency compared to the complex basis transformation and computation of NURBS.Results show that this proposed parametric model can represent Chinese calves with averaged error observed as low as 0.37%,far accurate than similar works by others.This work takes advantages of the excellent fitting ability and smoothness within data points of NURBS to provides an alternative solution for high-accuracy human calf modeling for customization of medical compression stockings.It’s suggested that the relative methods or technologies can facilitate high-accuracy design and fabrication of other wearable and compressive apparels for rehabilitation.