High spectral resolution mid-infrared hyperspectral computational ghost imaging enabled by wavelength-tunable nonlinear structured illumination

Leveraging atmospheric transparency and molecular fingerprint absorption in the mid-infrared (MIR) region, MIR hyperspectral imaging holds great promise for non-destructive material analysis, remote sensing, and microscopy. However, conventional MIR hyperspectral imaging systems suffer from the need for high-cost, cryogenically cooled multi-pixel detectors. Here, we propose a novel MIR hyperspectral computational ghost imaging scheme based on wavelength-tunable structured illumination, generated via difference frequency generation (DFG). To circumvent the lack of high-fidelity MIR spatial modulators, spatial patterns produced by a mature near-infrared spatial modulator are transferred to the wavelength-tunable, narrowband MIR light through the DFG process. As a proof-of-concept, we demonstrate MIR hyperspectral computational ghost imaging with a high spectral resolution of 1.73 cm−1 over the 3288–3408 nm range, using only a single-pixel MIR detector. Furthermore, real-time monitoring of dynamic methane (CH4) diffusion is achieved at 1.2 Hz under sub-Nyquist sampling. These results establish a practical route to MIR hyperspectral imaging that combines high spectral resolution, single-pixel simplicity, and fast data acquisition, offering significant potential for chemical discrimination and biomedical diagnostics.