The 3d-noise components are used for describing detector noise, especially for infrared focal-plane array (FPA)
detectors, where individual pixel variations and the read-out-process of pixel values will give rise to detector-induced
intensity variations. For consistency of actual and anticipated system performance it is essential that detector noise is
measured and estimated in a way that is consistent with how noise is modeled in system simulations. This paper
describes how to efficiently obtain bias-free estimates of the 3d-noise components in the frequency domain. Frequency
domain representation of the noise also allows application-tailored detector noise requirements. Intensity variations
caused by e.g. optics should not be allowed to influence the detector noise estimation, frequency domain estimation can
easily avoid influence from such non-uniformity. An alternative approach is to use pre-filtering as a means to suppress
non-uniformity; the effect of filter choice on estimated 3d-detector noise values is examined and filter choice discussed.
Finally, we demonstrate that low pixel operability, requiring replacement by interpolation from neighboring pixels for
many of the pixels in the array, will lead to under-estimation of the detector noise of the operable pixels.