Broadband Wide Field of View Imaging with Computational Mirrors
Single-sensor mirror system captures sharp images from visible to shortwave IR without refocusing.
Traditional glass-based optics struggle to focus across broad wavelength ranges like visible to shortwave infrared, limiting the potential of emerging VIS-SWIR sensors. While mirrors offer achromatic advantages, they suffer from field curvature and off-axis aberrations that degrade image quality over a wide field of view.
To solve this, Saragadam and colleagues developed Computational Mirrors, a hybrid approach that combines a simple concave mirror with a computational backend. By capturing just 2-4 focal stack images, the system uses a physics-inspired model called SeidelConv to characterize and correct spatially-varying point spread functions caused by off-axis aberrations. The team validated the framework with a 50mm F/1 system producing sharp images across 400–1700nm without refocusing, and a 100mm F/2 system for long-range imaging — opening new possibilities for compact, broadband imaging in defense, remote sensing, and material analysis.
- Uses only 2-4 focal stack images to computationally recover a sharp, all-in-focus image across VIS-SWIR.
- Introduces SeidelConv, a physics-inspired PSF model that corrects off-axis aberrations in simple concave mirrors.
- Demonstrated with 50mm F/1 and 100mm F/2 systems achieving sharp imaging from 400nm to 1700nm without refocusing.
Why It Matters
Enables compact, broadband imaging systems that reveal material details across visible, NIR, and SWIR bands.