Tech Briefs Magazine - November 2023 - PIT-17

Si LED integrated in a CMOS platform, "
SMART researchers described
their development of the smallest reported
Si emitter with a light intensity
comparable to that of state-of-the-art Si
emitters with much larger emission areas.
In a related breakthrough, SMART
researchers also unveiled their construction
of a novel, untrained deep
neural network architecture capable of
reconstructing images from a holographic
microscope in a paper titled,
" Simultaneous spectral recovery and
CMOS micro-LED holography with an
untrained deep neural network, " recently
published in the journal Optica.
The novel LED is a CMOS-integrated
sub-wavelength scale LED, which at
room temperature exhibits high spatial
intensity (102 ± 48 mW/cm2) and possesses
the smallest emission area
(0.09 ± 0.04 μm2) among all known Si
emitters in scientific literature. In order
to demonstrate a potential practical application,
the researchers then integrated
this LED into an in-line, centimeter-scale,
all-silicon holographic
microscope requiring no lens or pinhole,
integral to a field known as lensless
holography.
A commonly faced obstacle in lensless
holography is computational reconstruction
of the imaged object. Traditional
reconstruction methods require detailed
knowledge of the experimental setup for
accurate reconstruction and are sensitive
to difficult-to-control variables such
as optical aberrations, the presence of
noise, and the twin image problem.
The research team developed a deep
neural network architecture to improve
the quality of image reconstruction.
This novel untrained deep neural network
incorporates total variation regularization
for increased contrast and
takes into account the wide spectral
bandwidth of the source. Unlike traditional
methods of computational reconstruction
that require training data, this
neural network eliminates the need for
training by embedding a physics model
within the algorithm. In addition to holographic
image reconstruction, the
neural network also offers blind source
spectrum recovery from a single diffracted
intensity pattern, which marks a
groundbreaking departure from all previous
supervised learning techniques.
The untrained neural network demonstrated
in this study allows researchers to
use novel light sources without prior knowledge
of the source spectrum or beam profile,
such as the novel and smallest known
Si LED described above, fabricated via a
fully commercial, unmodified bulk CMOS
microelectronics process.
The researchers envision that this synergetic
combination of CMOS micro-LEDs
and the neural network can be
used in other computational imaging applications,
such as a compact microscope
for live-cell tracking or spectroscopic imaging
of biological tissues such as living
plants. This work also demonstrates the
feasibility of next-generation on-chip imaging
systems. Already, in-line holography
microscopes have been employed for a
variety of applications, including particle
tracking, environmental monitoring, biological
sample imaging, and metrology.
Further applications
include arraying
these LEDs in CMOS to generate programmable
coherent illumination for
more complex systems in the future.
INDUSTRIAL 3D DEPTH CAMERA
D400E SERIES
framos.com
Ruggedized, IP66 rated 3D camera
for industrial environments
State of the art stereo depth sensing camera.
Photonics & Imaging Technology, November 2023
17
http://info.hotims.com/84485-706
Tech Briefs Magazine - November 2023

Table of Contents for the Digital Edition of Tech Briefs Magazine - November 2023
