Summary
The goal of the DARPA XENA program is to develop new methods for image enhancement in long-standoff transmission X-ray scenarios where motion blur is also present.
Specifically, performers will deliver algorithmic toolsets that will be capable of creating useful inferences in terrestrial or aerial imaging scenarios where there is no prior information available about the interior composition of the object being imaged. XENA is focused on man-made objects, and methods that work for hard X-rays (≥150 keV).
If successful, the research conducted under XENA may create a new field: the exploitation and use of unresolved X-rays (as well as gamma rays, muons and other energies).
There are three major technical challenges XENA is addressing:
- Data sparsity due to Long range. Today’s state of the art for transmission X-ray is set by industrial or medical imaging transmission X-ray computed tomography (CT), which uses precisely calibrated equipment with exposure settings optimized for high contrast. XENA aims to push this state of the art out by at least three orders of magnitude – from single-meter ranges to single-kilometer ranges.
- Data sparsity due to motion blur. Motion blur is known to reduce signal-to-noise ratio (SNR). Therefore, new methods of processing long-standoff X-ray transmission imaging data are needed that include motion blur compensation.
- Lack of knowledge about the interior of the object being imaged. The majority of deblur, image sharpening, or image processing algorithms, particularly in the X-ray domain, rely on prior knowledge of the imaged object. XENA will focus on blind algorithm development with no prior knowledge.
The XENA Proposer’s Day presentation is shared for advance awareness only. The information shared in the XENA Proposer’s Day presentation is subject to change.
The XENA Proposer’s Day presentation does not constitute a request for abstracts/proposals, and it does not constitute a commitment of any kind by DARPA. Please monitor Special Notice (DARPA-SN-24-83) for additional information.