Filtration techniques are useful tools for analysing roundness profiles. The 2RC filter and Gaussian filter are commonly used to assess peripheral undulations of the roundness data. However they cannot do every aspect of functional prediction. Morphological filters are employed to characterize roundness profiles for functional assessment. Traditional computation methods for morphological filters are limited to planar surfaces and unable to be extended to roundness measurement. A novel method based on alpha shape theory is developed to break up the confinement. The morphological closing and opening envelopes are obtained by rolling a disk upon the roundness profile from the air and material side of the component respectively. They can be used to identify significant peaks and valleys on the profile respectively, which is vital to the functional performance of components, especially contact phenomenon. A case study is presented where various options of morphological filters and reference circles are applied to a roundness profile, delivering different functional meanings. An in-depth comparison of morphological filters and the Gaussian filter is followed to derive their pros and cons.