Compressive Imaging Unscrambles Light

A new compressive optical sensing technique conquers light scattering limitations of biological tissue, allowing a deeper look inside the body.

Developed by a team from Jaume I University (UJI) and the University of Valencia, a single-pixel optical system has shown that light scattering, and subsequent image scrambling that often occurs with existing technologies such as ultrasound and x-ray, can be averted when imaging the body’s tissues deeper than a millimeter under the skin.

An image of the Cheshire Cat sits in front of a scattering layer that completely hides it. The image was used to demonstrate the new single-pixel optical system’s effectiveness. Courtesy of Jesus Lancis/Jaume I University.

In the study, the researchers used a standard digital micromirror array from a commercial video projector to create a set of microstructured light patterns that are sequentially superimposed onto a sample.

The transmitted energy was measured with a photodetector that has no spatial resolution, according to the researchers, but can sense the presence or absence of light. The new single-pixel optical system, based on a compressive sensing signal processing technique, was then used to reconstruct the image.

“In compressive imaging, since we aren’t using pixelated sensors, it should be less sensitive to light scrambling and enable transmission of images through scattering,” said Jesus Lancis, a researcher in the Photonics Research Group at UJI and co-author of this study.

A standard multi-pixel imager cannot see the Cheshire Cat because the material in front scatters the light (a). The new single-pixel system overcomes any such limitations (b). Courtesy of Optics Express.

He added that this new technique could also operate through dynamic scattering, as calibration of the medium is not required, and “its fluctuations during the sensing stage don’t limit imaging ability.”

The research was published in Optics Express (doi: 10.1364/OE.22.016945).

For more information, visit ujiapps.uji.es.