Current signal post processing in spectrally encoded frequency domain (FD) optical coherence microscopy (OCM) and optical coherence tomography (OCT) uses Fourier transforms in combination with non-uniform resampling strategies to map the k-space data acquired by the spectrometer to spatial domain signals which are necessary for tomogram generation. We propose to use a filter bank (FB) framework for the remapping process. With our new approach, the spectrometer is modeled as a critically sampled analysis FB, whose outputs are quantized subband signals that constitute the k-space spectroscopic data. The optimal procedure to map this data to the spatial domain is via a suitably designed synthesis FB which has low complexity. FB theory additionally states that 1) it is possible to find a synthesis FB such that the overall system has the perfect reconstruction (PR) property; 2) any processing on critically sampled subband signals (as done in current schemes) results in aliasing artifacts. These perspectives are evaluated both theoretically and experimentally. We determine the analysis FB corresponding to our FD-OCM system by using a tunable laser and show that for our grating-based spectrometer - employing a CCD-line camera - the non-uniform resampling together with FFT indeed causes aliasing terms and depth dependent signal attenuation. Furthermore, we compute a finite impulse response based synthesis FB and assess the desired PR property by means of layered samples. The resulting images exhibit higher resolution and improved SNR compared to the common FFT-based approach. The potential of the proposed FB approach opens a new perspective also for other spectroscopic applications.© (2007) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.