BOE researched and developed 8K+AI ultra-high-definition video image technology solution: support automatic repair, HDR color adjustment algorithm

IT House reported on November 19 that classic film and television images carry the memories of a generation. However, as time flies, blurs, scratches, noise…have “dust” on the picture in memory. In recent years, the rapid development of artificial intelligence technology has brought new vitality to video works.

BOE Technology Innovation Center has developed and launched an 8K+AI ultra-high-definition technology solution, and based on a unique ultra-high-definition algorithm model, it effectively solves the problems of image clarity, color, and smoothness. It has been realized as a professional customer of Beijing TV Station, Guangdong TV Station, etc. Provide AI repair services that meet the ultra-high-definition standard, deeply empowering the ultra-high-definition display field.

BOE Technology Innovation Center’s 8K+AI ultra-high-definition technology has a wealth of technical accumulation in the field of image and video processing, and can be customized for different scene needs. With existing equipment, the automated process of BOE AI restoration can increase the efficiency of image restoration by 2-3 times and reduce labor costs by more than 50%.

BOE’s 8K+AI ultra-high-definition technology solution intelligently improves the image resolution, “one image, one set of parameters” adaptive monitoring, and uses 8K+AI ultra-high-definition technology to repair “old photos” with very powerful performance.

BOE stated that taking the restoration of historical photos of the “Forbidden City” as an example, some photos are more than 100 years old, with low resolution and uneven quality of each photo, which poses great challenges.

In response to these difficulties, the idea of ​​”first overall and then partial” was adopted, and the super-resolution technology of the BOE Technology Innovation Center was first used to adaptively stretch the resolution of the image to obtain a high-resolution image. Through the “one image, one set of parameters” method, the image processing process is adaptively monitored to solve the problem of not fixed parameters. Regarding face processing, the face will be located and targeted details will be improved.

See also  #stitch с @sarahjane.stevens Yeah, that's my unique talent #talent #sounds #c ...

IT Home is informed that, specifically, it uses a multiple reverse recursive deep network, which can multiply low-resolution images, and can achieve a maximum of 16 times stretching, which can be converted into high-resolution images from 4K to 8K. The picture is magnified without loss. At the same time, through GAN (antagonistic generative network, a machine learning architecture), it can intelligently supplement image pixel details and take into account the effects from macro to micro, making the generated images more realistic and natural.

The multiple reverse recursive deep network is an algorithm model based on deep learning. It learns the characteristic parameters of the image through hundreds of layers of brain neural networks. The stretched image is compared with the original input low-resolution image for feature comparison, which helps to enhance the accuracy of the stretched image and greatly reduce the loss of image features. It has the advantages of small system memory, fast image processing speed, and image stretching visual effects Good to wait for the advantages.

In addition, color enhancement technology has been able to intelligently color black and white photos, further shortening the distance between the audience and history.

Compression distortion repair technology + AI HDR color correction algorithm, so that the video repair effect is better. The effect of applying 8K+AI ultra-high-definition technology to improve video quality is also amazing.

When improving the video quality, if you follow the traditional way of thinking, simply brighten the dark part, it will cause the noise and compression block in the dark part to be more obvious. In this regard, BOE has adopted compression distortion repair technology to effectively suppress the noise interference in the dark part of the video.

See also  A powerhouse to remove car waste #fyp #foryou #goodthing

In addition, BOE designed an HDR intelligent color grading algorithm model. Through effective learning of colorists’ HDR grading experience, it forms a color grading model pool that can be adapted to different scenes, and adaptively performs HDR up-conversion and different styles of ultra-high-definition videos. Intelligent color creation. Through intelligent HDR color grading, ultra-high-definition video can be extended to the HDR standard and BT.2020 color gamut, and full use of the color gamut for color remapping, effectively improving the rich colors and levels of the video image, and improving the contrast of light and dark. And it makes the color expression more full, and the details are more abundant.

BOE Technology Innovation Center’s 8K+AI ultra-high-definition technology has also been successfully applied in the field of cultural and expo. On the one hand, 8K+AI technology can effectively restore the art material in ultra-high definition, making the material more plastic; on the other hand, through BOE’s excellent ultra-high-definition display products, the display of cultural relics is more humane and helps the audience to understand the cultural relics in depth. The cultural value behind.

At present, BOE has provided a wealth of digital display solutions for the museum industry, including the Palace Museum, to make artworks “live” and full of vitality.



Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.