3D printed decoder, AI-enabled image compression

AI Optical Decoder

Picture: The system makes use of an algorithm that encodes a high-resolution picture right into a lower-resolution one, after which interprets the compressed picture again to its authentic decision utilizing a decoder that decodes the incoming gentle.
view plus

Credit score: Ozcan Lab/UCLA

RECOMMENDATIONS

A UCLA workforce has developed a know-how to venture high-resolution computer-generated pictures utilizing one-sixteenth the variety of pixels contained of their supply pictures. The system compresses the photographs based mostly on a synthetic intelligence algorithm after which decodes them utilizing an optical decoder, a skinny, translucent sheet of plastic produced by a 3D printer, which is designed to work together with gentle in a selected means as a part of the identical algorithm. . The decoder doesn’t eat energy, which may end in larger decision shows that use much less energy and require much less knowledge than present show applied sciences.

BACKGROUND

Projecting high-resolution 3D holograms requires so many pixels that the duty is past the attain of as we speak’s shopper know-how. The power to compress picture knowledge and immediately decode compressed pictures utilizing a skinny, clear materials that doesn’t eat energy, as demonstrated within the research, may assist overcome that barrier and end in transportable know-how that produces higher-quality pictures utilizing much less energy and storage. than present shopper know-how.

METHOD

The system makes use of an algorithm that encodes a excessive decision picture to a decrease decision one. The result’s a pixelated sample, just like a QR code, that’s unreadable to the bare eye. That compressed picture is then translated again to its authentic decision by a decoder designed to bend and unscramble the incoming gentle.

By testing the system with pictures in black, white, and shades of grey, the researchers demonstrated that the know-how may successfully venture high-resolution pictures utilizing pictures encoded with solely about 6% of the unique pixels. The workforce additionally examined the same system that efficiently encoded and decoded shade pictures.

IMPACT

The know-how may ultimately be used for functions such because the projection of high-resolution holographic pictures for digital actuality or augmented actuality glasses. By encoding pictures utilizing a fraction of the info contained within the authentic and decoding it with out utilizing electrical energy, the system may result in smaller, cheaper holographic shows with sooner refresh charges.

The know-how may seem in shopper electronics as quickly as 5 years from now, in accordance with the paper’s corresponding writer, Aydogan Ozcan, Chancellor Professor of Electrical Engineering and Bioengineering, Volgenau Professor of Engineering Innovation on the UCLA Samueli College of Engineering and affiliate director of the California Institute for Nanosystems at UCLA.

Different potential functions embody picture encryption and medical imaging.

AUTHORS

Co-authors of the research are UCLA doctoral college students Γ‡ağatay Işıl and Deniz Mengu. Mona Jarrahi, the Northrop Grumman Professor of Electrical Engineering at UCLA, is a co-senior writer. Different authors are Yifan Zhao, Anika Tabassum, Jingxi Li, and Yi Luo, all from UCLA.

DAILY

The research is revealed in Advances of Science.

MONEY

The analysis was funded by the Division of Power.


Disclaimer: AAAS and Eurek Alert! should not liable for the accuracy of the press releases revealed on EurekAlert! by contributing establishments or for using any info via the EurekAlert system.

Leave a Comment