Compressive hyperspectral imaging reconstruction by spatial and spectral joint prior

Hyperspectral imaging systems can benefit from compressed sensing to reduce the size demand of sensor array. A new reconstruction algorithm is presented to recover the hyperspectral images from limited compressive measurements, exploiting the inherent spatial local smoothness prior, spatial nonlocal self-similarity prior and adjacent spectral similarity prior through joint regularization. The reconstruction process is solved with the help of augmented lagrangian multipliers and alternating direction method. The experimental results show that our method exhibits its superiority over other traditional methods with higher reconstruction quality at the same measurement rates.