Research On Scaling Of The Point Cloud Of Human Body Skin For Use In Digital Human Modeling

The modeling of the digital human body is a rapidly developing field in the study of ergonomics,with applications ranging from efficiency analysis to biomechanical calculations and computer animation.By creating an accurate representation of the human body,digital human body models can be used to precisely simulate how humans interact with products,environments,and technology.Oftentimes,it is desirable to create digital humans of different sizes,shapes,and body types,while using only a small number of simple anthropometric dimensions.However,constructing a digital human body model using detailed body size information can be a tedious and timeconsuming process.To address this issue,in this article an anthropometric dimensiondriven framework of human body skin modeling was proposed,which can be integrated into many digital human modeling systems to improve the efficiency and quality.The research framework of this paper mainly consists of three parts: establishing a database,establishing a body landmark prediction model,and scaling the human body skin point cloud.In order to establish a database,23 anthropometric dimensions were selected as user input parameters,and a series of representative points on the human body skin,called body landmarks,were determined as control points for the scaling algorithm.The selection of these body landmarks was grounded on earlier studies conducted by scholars,and was subsequently refined and fine-tuned based on the scaling effect.Meanwhile,human body size data samples were obtained through simulation,and corresponding human body models were constructed in CATIA software.Coordinate data for body landmarks were collected on each human body model,and a database of anthropometric dimensions and body landmarks was established.The predictive model of body landmarks is a crucial component of the research framework.Through discriminant analysis,the target anthropometric dimensions inputted by the users were grouped into distinct clusters,and a hierarchical estimation structure was established using a hierarchical estimation approach,thereby establishing a more accurate model for predicting anthropometric dimensions.To further predict body landmarks,principal component analysis was used to reduce the dimensionality of the original body landmark data,retaining only a small number of principal components with a large cumulative error proportion.Multiple regression analysis was then performed using the principal components as the dependent variable and the inputted anthropometric dimensions as the independent variable to establish a predictive model for the coordinate principal components of body landmarks.This enabled the calculation of body landmark principal components using the predicted anthropometric dimensions,and further restoration of the body landmark coordinates.The scaling of the human body skin point cloud was achieved through radial basis function interpolation.Firstly,a pre-existing human body skin point cloud was selected as the standard template,with body landmarks selected as the original scaling control points.During scaling,the body landmarks calculated using anthropometric dimensions were used as the target human body landmarks(target control points).In order to establish an ideal function relationship between the control points and achieve the goal of ideal scaling transformation for all points on the human body skin point cloud,three different scaling methods were studied,including global scaling,block scaling without transition,and block scaling with transition.The advantages and disadvantages of each method were compared,and a suitable method was ultimately selected.Drawing on the research discussed above,a comprehensive method was formulated in this article for scaling the human body skin,which is driven by anthropometric dimensions and a program has been coded in Python to validate the feasibility of the proposed approach.Compared to prior research,the approach presented in this paper incorporates real3 D human body skin scan point clouds as the basis for scaling,resulting in more lifelike skin shapes and enabling scaling at any time for newly scanned human body skin point clouds.Additionally,a block scaling modeling method has been proposed,which is fast,highly accurate,and widely applicable.