Non-Regularized Direct Superresolusion for Realistic Image Reconstruction

Researchers at the University of Arizona have developed a super-resolution (SR) technique that creates sharper images from low-quality or small cameras without fabricating the details. The novel technique applies motion to multiple images and then uses a non-regularized algorithm that directly solves a fully-characterized multi-shift imaging system to grant realistic restorations with physically accurate high resolution.

Background: 
Currently known SR techniques employ optimized priors and regularizers to deliver stable appealing restorations even though they deviate from physically accurate reconstructions. The popular inversion process creates detail that is not necessarily true, with ambiguous solution sets.

Advantages:

• Converts low-resolution images to high resolution without untrue artifacts
• Can work with any digital camera

Applications:

• Medical imaging
• Forensics

Status: Provisional Patent Application filed