High-precise compressive spectral imager based on comprehensive optical distortion model

Coded aperture snapshot spectral imager (CASSI) is a novel computational imaging technique to perform the fast acquisition of three-dimensional (3D) spatio-spectral information of target scene. However, the traditional CASSI imaging model fall short to address some important factors existing in the actual optical system, including the diffraction, aberration, vignetting, pixel mismatch, and non-uniform response over different spectral bands, thus greatly reducing the quality of reconstructed spectral images. In addition, the traditional calibration method for CASSI system is cumbersome and inefficient. To overcome these limitations, this paper proposes an optimization-based comprehensive optical distortion (OCOD) model to accurately depict the non-ideal effects in the physical CASSI system. The optimization-based method is used to calibrate the parameters in the proposed model to better fit the real measurement data. Using the proposed high-order model, the reconstruction quality of spectral images is effectively improved, and the tedious re-calibration step when changing the coded apertures can be avoided, thus promoting the usability of CASSI. A prototype of CASSI is built and the superiority of the proposed OCOD model is demonstrated by the simulation and experimental results.