Implementation and evaluation of spatio-spectral reconstruction for spatial coding compressive spectral imaging

Compressive spectral imaging combines traditional spectral imaging technique with compressive sensing and its preliminary application ability has been early explored by the usage on the occasions such as machine-vision and biomedical imaging. In this paper, spatio-spectral reconstruction based on sparse recovery method is presented and its performance is estimated. Simulating imaging experiment is carried out by using AVIRIS visible band data and specific module based on two-step iterative shrinkage/thresholding algorithm is built so as to execute reconstruction for the data cube. Statistical index is adopted so as to judge data fidelity quantitatively so that the effect of the regularizer index optimization and frame shooting number increasing on the improvement of reconstructed data fidelity can be distinctly revealed. The spatio-spectral featuring of different ground species are also evaluated by visual judgment on restored images and spectral curves.