Material detection by industrial infrared imaging spectrometer (IRIS) is a key technique in multiple industrial applications, including garbage collection, material analysis, and robot vision. However, IRIS often suffers from overlapped bands and random noises, which limit the precision of subsequent processing. In this article, we propose a novel Gabor transform-based mid-wave infrared (MWIR) spectrum restoration model by successfully exploring the intrinsic structure of the clean MWIR spectrum from the degraded one. First, a total variation-regularized Gabor coefficient adjustment descriptor is designed and incorporated into the spectrum restoration model. In addition to adjusting the Gabor coefficient distribution by total variation-norm, the L2-norm of gradient is leveraged to regulate the smoothness of the instrument degradation function. Then, the proposed model is inferred using an efficient optimization approach based on split Bregman iteration method and alternating minimization algorithm. Quantitative and qualitative experimental results demonstrate that the proposed model favorably outperforms the state-of-the-art approaches. The restored high-resolution MWIR spectrum can be used to rapidly detect different materials in intelligent video systems.