Photos

Researchers shrink high-resolution color cameras down to the size of a grain of salt: A review of digital photography


Researchers from Princeton University and the University of Washington have developed a high resolution color camera roughly the size of a grain of coarse salt.

This new sensor technology combines meta surface optics and machine learning models to reconstruct images through a nano-optical camera. Specifically, research paper detailing the technology says ‘Light-modulating nano-optical cameras at the sub-wavelength scale could enable new applications in fields ranging from robotics to medicine. Although supersurface optics offers a path to such microscopic images, existing methods have achieved much poorer image quality than cumbersome refractive alternatives, essentially limited by aberrations at large apertures and low f-numbers. In this work, we close this performance gap by introducing a set of neural nano-optical imaging,’

Previous micro-cameras (left) took pictures with low detail, chromatic aberration and distortion. The new system, neural nanooptics (right), produces sharper full-color images. Photos courtesy of researchers.

The camera is based on a technology called hypersurface, which consists of 1.6 million cylinders. Each post is roughly the size of human immunodeficiency virus (HIV). Each post has unique geometry and functions like an optical antenna. According to Princeton, ‘Changing the design of each column is necessary to precisely shape the entire optical wavefront.’

Machine learning-based algorithms turn the lighting information from each post into actual images. Furthermore, the image quality surpasses anything that previous ultra-compact cameras could achieve. ‘A key innovation in camera making is the integrated design of the optical surface and the signal processing algorithms that generate the images. This has boosted the camera’s performance in natural light, in contrast to previous supersurface cameras that required lab-pure laser light or other ideal conditions to produce images. high quality photo ”, Felix Heide, who research by senior author and assistant professor of computer science at Princeton.

‘Our learned ultrathin superphotometer as in (a) has a thickness and diameter of 500 μm, allowing for the design of a miniature camera. Optically produced is shown in (b). Magnification is shown in (c) and the nano column size is shown in (d). Our end-to-end imaging pipeline presented in e includes a proposed efficient hypersurface image formation model and a feature-based decoding algorithm. From the optimized phase profile, our discriminable model generates spatially different PSFs, which are then patch-modified with the input image to form sensor measurement. The sensor reading is then decoded using our algorithm to produce the final image. The illustrations above “Meta-Optic” and “Sensor” in (e) were created by the authors using Adobe Illustrator. ‘

Image credits and captions: Ethan Tseng, Shane Colburn, James Whitehead, Luocheng Huang, Seung-Hwan Baek, Arka Majumdar & Felix Heide / Princeton University & University of Washington

Previous micro-sized cameras captured blurry, distorted images. New nano-optical technology produces better, sharper images with more accurate colors and an expanded field of view. Computer science PhD student Ethan Tseng said: ‘It’s quite a challenge to design and configure these tiny nanostructures to do what you want. student at Princeton who co-led ‘tudy. ‘For the specific task of large-field RGB imaging, it was previously unclear how to co-engineer millions of nanostructures along with post-processing digital spans.’

Co-author Shane Colburn, Ph.D. student at the University of Washington’s Department of Electrical and Computer Engineering, solved this problem by creating a computational simulation to automatically test different nanoantenna configurations. Colburn is currently an associate professor at the University of Washington.

‘Compared with existing state-of-the-art designs, the proposed neural nanooptics produces high-quality wide FOV reconstructions corrected for aberrations. Sample reconstructions are shown for a still life with fruits in (One), a clear green lizard (NS), and a clear blue flower (NS). Content is displayed below each row. We compare our reconstruction with basic truth-gathering using high-quality six-element compound refractometers, and we demonstrate that the reconstruction is accurate despite the mass of Our supermarket is 550,000 times lower than photosynthetic. ‘

Image credits and captions: Ethan Tseng, Shane Colburn, James Whitehead, Luocheng Huang, Seung-Hwan Baek, Arka Majumdar & Felix Heide / Princeton University & University of Washington

Student research and co-author James Whitehead, fabricated a silicon nitride-based supersurface. According to research, the supersurface design can be mass-produced at a lower cost than traditional in-camera lenses.

The team’s approach itself is not new. However, combining surface optical technology with neural-based processing is. Microcameras could have significant uses in medical settings to enable minimally invasive endoscopy. It can also improve imaging capabilities for robots with size and weight constraints. Possibly thousands of tiny cameras could be placed in an array, turning a surface into a camera.

The study can be read in full here. Its authors include Ethan Tseng, Shane Colburn, James Whitehead, Luocheng Huang, Seung-Hwan Baek, Arka Majumdar, and Felix Heide.



Source link

news7g

News7g: Update the world's latest breaking news online of the day, breaking news, politics, society today, international mainstream news .Updated news 24/7: Entertainment, Sports...at the World everyday world. Hot news, images, video clips that are updated quickly and reliably

Related Articles

Back to top button