AI Upscaling Technology

The 8K resolution display shows clearer images closer to analog than 4K resolution, but there is still a lot of SD content. Content and display of various resolutions will coexist.

It is common for low-resolution images to be input to high-resolution displays, and upscale technology is inevitable. As a result, interest in upscale technology is increasing.  Recently, learning-based upscale technology is increasingly achieving high performance.

We will look at the trends and strengths and weaknesses of various upscale technologies in terms of TV upscale, and explain Samsung’s AI upscale technology.


Seungho (SH) Park, Principal Engineer, Samsung Electronics Video Display Business

SH Park was a researcher at the Samsung DMC Research Center for about 10 years from 2005 to 2014 and was a visiting researcher at the Stanford Computer Forum for six months in 2012. He has been working in Visual Display for five years from 2015. Currently he is a principle engineer and is in charge of image quality algorithms in the Visual Display division.  His work experience includes H.264 codec implementation, frame rate conversion, contrast enhancement, HDR development and standardization, resolution upscaling.