Digital super-resolution for TDI remote sensing via micro-misaligned pushbroom imaging and computational reconstruction

Although super-resolution reconstruction algorithms are effective for increasing the amount of real information obtained, applying them to the commonly used the time delay integration architecture in spatial systems remains challenging, resulting in significant additional costs or unsatisfactory super-resolution performance. To address this issue, this paper presents an intelligent micro-misaligned pushbroom imaging architecture, which involves a trade-off between slight motion blur degradation and subpixel sampling. High-spatial-resolution information is reconstructed with an equivalent time–space bandwidth product without any additional optoelectronic hardware or motion mechanisms in space systems. Furthermore, we propose a physical model for rigorous digital super-resolution based on the imaging architecture. A novel sparse regularization term is proposed to increase model robustness. Various numerical methods are applied to significantly reduce the computational cost of the algorithm. The experimental results show that the proposed method can achieve up to 2 × digital super-resolution with approximately 1% reconstruction error. Compared with previous methods, our method provides more real details and clearer reconstruction images with lower errors in actual space environments. The proposed approach addresses the bottleneck associated with the difficulty of manufacturing photodetectors with small pixel sizes due to fabrication errors in remote sensing systems with submeter ground sampling distances. The data, codes, and models are publicly available at https://github.com/LushOrchid/SR-MMP .