Princeton University shrinks the camera to the size of a grain of salt

Now, researchers at Princeton University and the University of Washington have overcome these obstacles with an ultra-compact camera the size of a coarse salt grain

Through the joint design of camera hardware and computing processing, the system can realize the minimally invasive endoscopic diagnosis and treatment of diseases of medical robots, and improve the imaging size and weight limitations of other robots

Traditional cameras use a series of curved glass or plastic lenses to bend light into focus, while the new optical system relies on a technology called metasurface, which can be produced like a computer chip

Each pillar has a unique geometric shape, which functions like an optical antenna

A key innovation created by the camera is the integrated design of the optical surface and the signal processing algorithm that produces the image

The researchers compared the images produced by their system with the results of previous metasurface cameras.

Other ultra-compact metasurface lenses have severe image distortion, a small field of view, and limited ability to capture full-spectrum visible light-called RGB imaging because it combines red, green, and blue to produce different hues

"Designing and configuring these tiny microstructures to do what you want is a challenge," said Ethan Tseng, a PhD student in computer science at Princeton University, who is one of the leaders of the research.

Co-lead author Shane Colburn solved this challenge by creating a computational simulator to automatically test different nano-antenna configurations

As a doctoral student in the Department of Electronic and Computer Engineering (UW ECE) at the University of Washington, he presided over this work, and he is now an assistant professor in the school

Co-author James Whitehead is a PhD student at the European School of Economics.

"Although optical design methods are not new, this is the first system to use surface optics for front-end and neural-based processing," said Joseph Mait, Mait-Optik consultant, former senior researcher, and chief scientist at the US Army Research Laboratory

"The significance of this work lies in the completion of a difficult task, that is, the joint design of the size, shape and position of millions of features on the metasurface, as well as the parameters of post-detection processing, in order to achieve the expected imaging performance," did not participate in this.

Hyde and his colleagues are now working on adding more computing power to the camera itself

Heide also envisions using ultra-small imagers to create "sensor surfaces

In addition to Zeng, Cobain, Whitehead, Maqiangda, and Hyde, the authors of this study include Huang Luocheng, a doctoral student at the University of Washington; and Seung-Hwan Baek, a postdoctoral researcher at Princeton University

This work was partially supported by the National Science Foundation, the U.
S.
Department of Defense, the University of Washington Reality Lab, Facebook, Google, Futurewei Technologies, and Amazon
.

Journal Reference :

1. Ethan Tseng, Shane Colburn, James Whitehead, Luocheng Huang, Seung-Hwan Baek, Arka Majumdar, Felix Heide.
   Neural nano-optics for high-quality thin lens imaging .
   Nature Communications , 2021; 12 (1) DOI: 10.
   1038/s41467- 021-26443-0