Removing Impulse Noise from Human Head Scan Data

摘要

This report presents a mathematical morphological filtering technique toeliminate impulse noise from human head surface data generated by laser surface scanning technologies. In the area of image processing, mathematical morphology refers-to a branch of nonlinear filters which use geometric form and structure to alter a signal. Several structuring elements were investigated with a spherical structuring element being selected as the best candidate. A spherical element is the ideal shape since protruding shapes like spikes on the human head are best removed using a structuring element that has an omni-directional behavior. A spherical structuring element having a radius of 3.Omm to 3.5mm represented the best trade-off between identification of impulse noise and loss of good data points. (AN).