Fourier and Hadamard transform spectrometers: a limited comparison.

We establish an encoding figure of merit for a detector-noise limited Fourier transform spectrometer (FTS) and compare it to the comparable figure for a Hadamard transform spectrometer (HTS). If N measurements are made to establish N spectral densities, the mean square errors obtained with the Fourier system are a factor of 2 greater than for the analogous Hadamard system. The limitation of the Fourier system is partly that it does not truly Fourier analyze the radiation. Instead a cosine squared modulation is imposed on the different spectral frequencies. An additional difficulty is that neither the cosine nor the cosine squared functions form an orthonormal set. This makes the Fellgett's advantage (root-mean-squared figure of merit) for a single detector Michelson interferometer a factor of (N/8)(1/2) greater than for a conventional grating instrument-rather than (N/2)(1/2) as maintained in standard texts. The theoretical limit, which may not be realizable with practical instruments, would be (N)(1/2).