BETTERView offers SD to true HD (High Definition) conversion and video enrichment. Check the following video to see the quality advantage.
The patented breakthrough in the super resolution promises to breaks the “glass ceiling” of existing technology, says the company, whose team includes world-experts from Technion’s Faculty of Computer Science, which is ranked #15 in the world. While conventional methods use conversion techniques to “blow up” or stretch the SD video onto an HD display, BetterVirw increases optical resolution of a video stream, generating an HD stream that looks as authentic as it gets.
BETTERview technology is based on a novel family of SR algorithms, proposed by a world-leader in this field, Prof. Michael Elad (Technion – Israel Institute of Technology). Elad and his collaborator, Dr. Matan Protter devised the first method that overcomes the requirement for very accurate and explicit motion estimation in previous SR technologies. The new family of SR techniques avoids the exact motion estimation and replaces it by a probabilistic estimate. This enables handling successfully general content scenes containing extremely complex motion patterns.
The results are impressive, with no visual artifacts, and the process is completely robust. Based on this core technology, BETTERview developed the first cutting-edge industrial-grade robust system that perform SD to True HD resolution conversion. Its solution strengthens the above-mentioned core technology by handling various video artifacts, interlaced content, synchronization issues and run-time efficiency.
The innovative research of Prof. Michael Elad was listed in 2010 by Thomson Reuters Science Watch. You can read their interview with him here.
It has been known for the past 20 years that, in principle, one could take several low-quality images and fuse them into a single, higher-resolution outcome. This has been demonstrated by scientists, adopting various techniques and algorithms. The process is known as Super-Resolution (SR), which became a hot field in image processing, with thousands of academic papers published during the past two decades on the problem and ways to handle it. The classical approach to fuse the low-quality images requires finding an exact correspondence between their pixels, a process known as “motion estimation”. Several years ago this field experienced a revolution, due to a breakthrough in the way to handle (or better yet, bypass altogether) the motion estimation.