Funded by the Defense Advanced Research Projects Agency (DARPA), COMP-I is an acronym for Compressive Optical MONTAGE Photography Initiative (MONTAGE is short for Multiple Optical Non-redundant Aperture Generalized Sensors). The goal of this project is to reduce dramatically the volume and form factor of visible and infrared imaging systems without loss of resolution. The approach is to use an array of small lenses with extremely short focal lengths instead of a single conventional lens with large diameter and focal length. The array of lenses will create multiple copies of the scene on a single focal plane, and the challenge is to reconstruct an image with resolution comparable or higher to that of a conventional system. In order to achieve this each sub-aperture must be uniquely coded in order to ensure the images are non-redundant. Multiple coding strategies are being explored and developed. Some designs include the use of optical elements ranging from absorptive masks to birefringent crystals and even holographic gratings.
The project has envisioned its developments unfolding in two phases:
- Phase I involves the creation of a megapixel monochrome visible camera thinner than 2.5 mm front with a 5 mm lenslet aperture to focal plane surface and LWIR 120 by 160 imaging system. In this stage the visible conformal wide field of view imaging systems will make use of both fiber and holographic mapping.
- Phase II will lead to the development of a megapixel color camera thinner than 2.5 mm using compressive sampling and coded color masks. Furthermore, we will engineer a monochrome megapixel visible camera thinner than 1 mm coding aperture to focal plane, subject to successful holographic and photonic crystal lens fan-out elements, This phase will also see a LWIR flexible data cube sampling system thinner than 5 mm, capable of 1024 by 1024 imaging, dynamically remappable as spatio-spectral, multiresolution or spatio- temporal sampling. The final goal of this phase will be a conformal multiple aperture system mounted on a Army-relevant platform and delivered to NVESD.
One of the most effective advances in the project has been the development of compressing sampling. Compressive sampling refers to signal reconstruction using less than one measurement per estimated signal value. Compressive sampling is enabled by multiplex sensing, under which measurements are dependent on multiple points in the signal space, and signal inference, which is used to reconstruct the signal from multiplex data. An experimental setup is being developed in our laboratory to characterize and analyze quantitatively system performance as seen in the pictures below.
Testing the Calibration
Distortion Correction of Unknown Source
The COMP-I program represents a collaborative team approach to research and development. Working with Duke University's DISP group are the leading manufacturer of compact IR focal planes, the leading micro-optical assembly manufacturer, and the only academic groups to have demonstrated efficient computational sensors and infrared fan-out and sampling optics. More information about our partners can be found on their websites: