Performance of digital filtering methods in orthotic and prosthetic CAD/CAM.

Abstract

This report characterizes the performance of three digital filters, when applied to residual limb shape maps. The three filters were an averaging filter, a uniform window Fourier filter, and a Hamming window Fourier filter. The frequency responses of the three filters were calculated from theory, and experimentally observed. Experimental observations consisted of responses on single-frequency lobed shapes, and on residual limb shapes. Seven trans-tibial limb molds were digitized, three times each. Each resulting shape was then passed through each of the three filters. The before and after shapes were then compared. A Hamming window filter (low-pass frequency of 10 cycles per revolution, 24 coefficients, Hamming number of 0.25) achieved the best performance based on maintained amount of residual limb frequencies, and on visual observation.