Channeled spectropolarimeter (CSP) measures the spectrally resolved Stokes vector of light from only one single spectralacquisition, which makes it possible to accurately measure dynamic events. The accurate reconstruction of Stokes vectorplays a key role in this snapshot technique shifting the main burden of measurement to computational work. The state-ofthe-art algorithm runs the Fourier transform of the channeled spectrum or linear operator model of the system and itspseudo-inverse to reconstruct Stokes vector. However, they may suffer from the lack of signal-to-noise ratio (SNR) thenreduce the accuracy of reconstruction. To accurately reconstruct Stokes vector from noise-contaminated data, we proposean effective method called fast compressed channeled spectropolarimeter (FCCSP). In our FCCSP method, the spectrumfrom spectrometer is seen as the compressive representation of Stokes vector, thus the FCCSP algorithm is to solve anunderdetermined problem, where we reconstruct the 4N×1 Stokes vector from only N×1 spectral data acquisition points.Simulation results show that our FCCSP method is more accurate to reconstruct Stokes vector changing gradually withwavelength from noise-contaminated spectrum than Fourier and linear operator methods. Besides, it is faster and morememory and computation-friendly than other compressed CSP method.
展开▼