LAUSR

Abstract In this paper, a multispectral imaging method using compressed Fourier spectrum is introduced. It has similar optical architecture with the Hadamard Transform Spectral Imaging (HTSI) method, but needs less number of measurements for multispectral image reconstruction. The key is encoding the spectra by sinusoid coding patterns generated by Digital Micro-mirror Device (DMD). Sinusoid patterns transform the spectra into Fourier domain. The multispectral images can be reconstructed using finite Fourier coefficients by Inverse Fourier Transform (IFT). A testbed is built and two sets of experiments are conducted. The experiment results show that, about 10% of Fourier coefficients would be able to reconstruct the multispectral images well, and the number of measurements is about 80% less than that of the HTSI. Additionally, the presented method provides an alternative way to realize Fourier Transform Spectral Imaging (FTSI) by use of grating and DMD, distinguished from the interferometer-based FTSI and the sinusoid filter-based FTSI.