Computational Imaging

Computational imaging refers to the enhancement of the capabilities of an imaging system by the application of computational techniques, as an alternative to increasing the optical setup complexity. One of the aims of computational imaging is the development of cost-effective imaging systems, supported by the continuous price reduction of computation power predicted by Moore law.

Some everyday examples of computational imaging techniques, present in many commercial cameras, are computational increased field of view (panoramic pictures), or computational increased of the high dynamic range (HDRI).

Other computational imaging techniques have been source of intense research in the recent years, such as:

• Synthetic aperture
• Light field / Integral imaging.
• Pixel super-resolution.
• Computational multi-spectral imaging.
• Computational illumination.

In the Imaging Concepts group we have developed and demonstrated, for the first in thermal infrared (LWIR band), a computational imaging system based on a multi-camera array with simultaneous computational super-resolution, and light field/integral imaging. By timely making the most of the recent price reduction in this infrared imaging technology, we have built a 6x camera array of FLIR Lepton detectors, able to perform a digital refocus with a simultaneous resolution enhancement toward the diffraction limit of the camera, effectively doubling the resolution, and getting completely rid of the aliasing artefacts, while adding light field/integral imaging capabilities like robust 3D imaging in complex and dense scenarios, even through obscuration, in a tomographic-like way.