Mesh denoising in prosthetics manufacturing applications using average filtering, linear heat diffusion and bilateral filtering

Abstract

It is important for every person using a prosthesis that the prosthetic socket fits snugly to the extremity stump. The authors propose to improve the preprocessing stage of a three-dimensional scan extremity stump model using denoising. Ten 3D scans were selected for testing. Additive white Gaussian noise was considered. Three denoising algorithms (average filtering, linear heat diffusion, bilateral filtering) were discussed. SNR was calculated before and after denoising. Then the δ-SNR and the δavg-SNR were calculated. The execution time of algorithms with averaging more than 1000 runs was also estimated. Denoising was performed on three-dimensional models obtained by optical scanning of patients with lower limb stumps. The results of the work help solve the pretreatment problem for a stump model to an individual prosthetic socket manufacture.