Search

The basic principle of Fourier-transform imaging spectrometer (FTIS) based on Savart interferometer is outlined. The un-identical distribution of the optical path difference which leads to the wavelength drift of each row of the interferogram is analyzed. Two typical methods for wavelength calibration of the presented system are described. The first method unifies different spectral intervals and maximum spectral frequencies of each row by a reference monochromatic light with known wavelength, and the dispersion compensation of Savart interferometer is also involved. The second approach is based on the least square fitting which builds the functional relation between recovered wavelength, row number and calibrated wavelength by concise equations. The effectiveness of the two methods is experimentally demonstrated with monochromatic lights and mixed light source across the detecting band of the system, and the results indicate that the first method has higher precision and the mean root-mean-square error of the recovered wavelengths is significantly reduced from 19.896 nm to 1.353 nm, while the second method is more convenient to implement and also has good precision of 2.709 nm. [ABSTRACT FROM AUTHOR]

The basic principle of channeled Fourier-transform imaging spectropolarimeter (CFTISP) is outlined. The two mainstream techniques existing for performing polarization state demodulation are analyzed, which show uncertainty that may not be suitable for CFTISP based on lateral shear interferometer. A modified demodulation method for Stokes parameters is described. The method separate the phase of the sign and the high-order retarders’ retardations from the total phase acquired from the fast Fourier transform of the interferogram, which will not cause the amplitude error from the reference beam. Furthermore, the retardations and the residual phase error in each band introduced by instrument can be seen directly in this method. The effectiveness of this method is experimentally demonstrated with four known input states of polarization, and the results are satisfactory. The RMS error of each Stokes parameters is also presented, which demonstrates that the low spectral signal-to-noise ratio can increase the RMS error by nearly a factor of 2–5 for the individual Stokes parameters. The comparison of reconstructed results by four methods further demonstrates the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]