Google reveals RAISR: an image enhancement tech which uses machine learning

Google has shared details on a new image enhancement technique it has developed called RAISR. RAISR, or rapid and accurate image super-resolution, incorporates machine learning to produce high-quality versions of low-quality images.

Google Research Scientist Peyman Milanfar explained the technology on the Google research blog and how it differs from existing methods image enhancement methods. While Upsampling, a common way to increase the quality of low-resolution images, relies on fixed values – applying the same effect to every image, RAISR uses a non-linear approach.

With RAISR, the team employs machine learning to “train on pairs of images, one low quality, one high, to find filters that, when applied to selectively to each pixel of the low-res image, will recreate details that are of comparable quality to the original,” said Milanfar.

Essentially, RAISR uses high-quality images as a benchmark from which its algorithm can establish ways to adjust a low-resolution image. See its effect in the photos below (RAISR’s image enhancement is applied to the bottom one.)

There’s no indication of how Google might use the technology commercially but Milanfar says that it can be run “on a typical mobile device in real-time.” It seems possible that RAISR could be integrated into Google’s Camera app or an upcoming smartphone.