On-Orbit SWIR Imagery MTF Compensation for the Geology-1 Satellite Using Cross-Spectral Structural Priors

The Geology-1 hyperspectral satellite employs a lightweight and highly integrated payload architecture, establishing a novel paradigm for commercial small-satellite-based geological observation. However, the compact opto-mechanical configuration exacerbates the coupling among residual attitude disturbances, cryocooler-induced micro-vibrations, and the imaging chain, leading to systematic degradation of the system Modulation Transfer Function (MTF) in shortwave infrared (SWIR) imagery. To address this challenge, we propose a structure-guided, two-stage Geology-1 SWIR Imagery MTF Compensation (GSMC) framework. This framework leverages co-registered visible and near-infrared (VNIR) imagery, which offers superior spatial resolution and enhanced imaging stability, as a natural structural prior. Consequently, a zero-shot compensation model is constructed without requiring external training data. Experimental results demonstrate that GSMC delivers consistent and superior performance across diverse geological scenes. Quantitative evaluation shows that the MTF at the Nyquist frequency increases from 0.204 to 0.297 after compensation. The proposed method preserves high radiometric fidelity, with a mean relative error of only 0.57%, while also improving downstream task performance by increasing land-cover classification accuracy from 83.43% to 89.19%. In addition, spectral consistency analysis verifies that the cross-spectral guidance strategy does not distort SWIR-specific information, with spectral angle mapper (SAM) values remaining below 0.05. These results indicate that the proposed framework effectively compensates for time-varying jitter degradation in Geology-1 imagery and provides reliable technical support for quantitative on-orbit processing and high-precision geological mapping.