In the field of optical metrology, there are many existing techniques for measurement of the surface profile or film thickness of a sample of interest. However, many of these common techniques break down when there exists a high level of noise in the output optical signal due to the nature of certain types of samples. Sources of noise that can suppress the signal information can come from highly rough surfaces and turbulence within the medium if there is necessity for light to transmit through. This dissertation presents the results of a research effort to develop a technique to suppress the noise content in order to distinguish the signal. In particular, a noise-reduction technique, Fractional BiSpectrum, was developed for the purpose of properly identifying signal frequencies within the measurement data when these frequencies have a known relationship with each other. It is also shown how this technique can be implemented within common interferometry setups such as those measuring the thickness of films with turbid material as well as the surface profile and/or roughness of a sample and still make the proper measurement, i.e. identify the proper temporal or spatial frequency within a noisy signal. It is demonstrated that using this noise-reduction technique enables to produce a confident measurement – i.e. enhance SNR greater than 3dB. In some of the demonstrated cases, enhancement will lead to the detection of the correct measurement parameter – i.e. thickness or height value - where the existing techniques fail.