Snapshot compressed sensing computed-tomography imaging spectrometry

Abstract

Computed-tomography imaging spectrometry (CTIS) can achieve non-scanning and high speed imaging recording of spatial and spectral data of a rapidly changing target scene. However, it has the problem of missing cone, which means the projection data cannot be fully sampled. This limits its practicality due to the ill-posed spectral reconstruction from limited angles of projection tomography. This paper proposes a compressed sensing (CS) sampling model for the CTIS, or CSCTIS in short, with the under sampling advantage of CS to improve the problem. The simulation results validates that the CS model is more effective than the traditional computed-tomography (CT) one, and further experimental results prove that the CSCTIS performs more accurate spectral reconstruction than the traditional CTIS.