Correction for a long-wave infrared coded aperture hyperspectral imaging model via multi-source deviation coupling analysis

The coded aperture hyperspectral imaging systems typically describe the imaging process using a decoupled "ideal coding model plus error" approach. However, in the long-wave infrared band, non-ideal coding of the mask component and stray light from unintended grating orders are deeply coupled with the encoding-dispersion process, rendering traditional models ineffective. This paper proposes a coupled corrected imaging model that accounts for both interferences and further considers their mutual coupling. A corresponding reconstruction strategy is then developed, which, combined with conventional estimation algorithms, enables full-scene data recovery. A laboratory gas detection experiment was conducted to compare the reconstruction performance of the traditional model, the partially corrected model, and the fully corrected model. Results demonstrate that, compared to traditional models, the proposed model effectively eliminates structured errors caused by these interferences: spatial striping artifacts are removed, the characteristic absorption peaks of ammonia are correctly reconstructed, and no fake peaks are introduced in the weak absorption bands, validating the effectiveness of the proposed model.