Finland-based company Visidon, founded in 2006, specialized in the development of AI-based image and video enhancement software. The company is a well-known supplier for the top mobile OEMs around the world. Lately, the company has utilized its long experience and has further developed its technologies to serve the broadcasting and video delivery industry.

New video enhancement solutions to bring archived video content up to date

“We have been focusing on developing software to improve the overall visual quality of archived video streams. With our long experience from optimizing algorithms to embedded platforms, we are well known for delivering technologies that are optimized both in accuracy and computation performance,” says Visidon CEO, Markus Turtinen.

The main challenges, that Visidon´s clients want to solve are related to low visual quality of video streams including interlacing, low resolution, noise, artifacts, and low frame rate. The new technologies for video enhancement include for example super resolution, frame interpolation, deinterlacing, and noise reduction. Super resolution refers to AI-based upscaling from lower resolutions to 4K or 8K, which makes videos streams compatible with modern displays. It also increases details and clarity in the video. Frame interpolation increases the frame rate and makes especially fast movement, like sport content, look smoother. It also allows creation of slow-motion clips. Deinterlacing converts interlaced videos to progressive scan, and noise reduction enhances the quality of noisy low-light video streams.

According to Visidon, the long experience and strong background in visual quality improvement and computer vision applications makes the company stand out from the competition. Being at the forefront in delivering superior imaging and video enhancement quality for hundreds of millions smartphone camera users during the past years, means that Visidon has gained unrivalled experience in creating high performing software products. Limited computational power on battery-based devices has driven Visidon to develop fast and energy efficient solutions. Also, in the video industry there is high demand for processing a massive amount of video streams efficiently. Bringing exceptional visual quality and restoring videos at a reasonable hardware cost will be important in scaling to the huge amount of video contents that is broadcast and streamed today and in the future.

Broadcasters and content delivery providers utilizing AI to restore old videos streams

Broadcasters worldwide are looking for ways to differentiate themselves. They have large libraries of archived video content that could be utilized to generate more revenue streams and to widen their content portfolio, as well as to attract a wider audience. Visidon has already been collaborating with broadcasters from different continents to bring their old archive video content up to date. This is done with technologies such as super resolution and de-interlacing.

Most current domestic TV´s support 4K resolution and viewers are even expecting to see their favorite childhood series and movies in 4K format but of course most of these are still in SD quality. And here comes the problem: SD quality translates into a resolution of 720x576 pixels at best, which is only 5% of the pixel count of a 4K screen. This means that a standard 4K TV is converting the stream by enlarging each frame 20 times in order to fill the entire 4K screen. This usually results in a blurry image. If the video source file is processed with Visidon Super resolution, an AI based neural networks does the upscale and instead of enlarging each frame, AI adds realistic detail to the 19 newly generated pixels for each pixel of the original video and the final 4K result has a lot more details than the conversion done by traditional scaling methods used in display hardware. A lot of the old video content is also interlaced, and Visidon has developed their own deinterlacing technology for smart conversion into progressive format providing a significantly better outcome compared to industry standard approaches available - for example in ffmpeg.

Another option for upscaling the content would be traditional remastering, which means scanning the original film digitally frame by frame. Compared to relatively fast process of AI-based super resolution utilizing neural networks, this is extremely slow and can rarely be justified from a business perspective.

Visidon one of the newest members of IABM

Visidon has just joined the IABM network and is more than happy to receive inquiries around visual quality enhancement of videos. Visidon is offering flexible licensing and business models and as a member of the IABM network, the company aims to be even better informed and connected with the media and video content delivery industry in order to further develop our solutions to enable our clients to provide better video quality and greater viewer experience.

Read more: www.visidon.fi