Research On Personalized Prosthetic 3D Printing Modeling Methods

Prosthetics are an important medium to ensure the normal life of disabled patients,and the traditional prosthesis production process is complex,long and costly,which is difficult to meet the needs of rapid production of personalized prosthetics.The 3D printing technology can optimize the prosthetic manufacturing process and provide new ideas for personalized production.In the personalized prosthetic production,the accuracy of the stump point cloud model directly determines the quality of prosthesis production.However,the modeling process of the stump point cloud model has many problems such as noise and high density,and the existing algorithms are difficult to accurately extract the detailed features of the stump model,resulting in low accuracy and poor comfort of the stump model.In this paper,the personalized prosthetic 3D printing reconstruction technology is taken as the research background,and the lower limb stump is taken as the research object.In order to improve the manufacturing accuracy of personalized prosthesis,the stump point cloud filtering,simplification and reconstruction technology are studied.The main research work is as follows:(1)Research on stump point cloud filtering algorithm.In order to improve the noise recognition ability of the residual limb point cloud,improve the adaptability of the filtering method,and solve the problem of different types of noise affecting the reconstruction accuracy in the residual limb point cloud,an improved adaptive combined filtering algorithm for the residual limb point cloud is proposed.An adaptive standard deviation multiple based on the volume of the local point cloud is introduced to design an adaptive distance threshold scheme to filter out the large-scale noise of the residual point cloud.The weighted covariance matrix is used to improve the accuracy of the point cloud normal vector,and the feature weight factor is optimized by the mean normal angle change degree,the feature retention of the bilateral filter factor is improved,and the small-scale noise of the residual limb point cloud is smoothed.(2)Research on stump point cloud optimization algorithm.In order to solve the problem that the data density of the stump point cloud after filtering and the holes in the non-characteristic area of the stump are caused easily by the curvature simplification algorithm,a stump point cloud reduction algorithm based on curvature grading is proposed.The curvature simplification algorithm was used to extract the stump feature point cloud.The bisecting K-means clustering method is introduced to divide the non-characteristic regions of the stump point cloud,and the noncharacteristic areas of the stump point cloud are simplified sequentially by the curvature simplification method,and the stump feature point cloud is output.(3)Research on stump point cloud reconstruction algorithm.Aiming at the problems of holes,redundant triangular patches and insufficient expression of detailed features in the stump model,a 3Dprosthetic reconstruction scheme based on attention mechanism is proposed.The feature embedding layer is used to realize the transformation of the stump point cloud and the feature map.The edge convolution feature extraction module is used to extract local details,and the feature fusion module is used to aggregate the stump point cloud features to enrich the stump shape information.The implicit decoder is used to determine the position relationship between the stump point cloud and the shape,and the iso-surface of the stump model is extracted by the moving cube method.