Image super-resolution reconstruction method and device for real-time video stream

Abstract

The invention discloses an image super-resolution reconstruction method and device for a real-time video stream, and the method comprises the steps: obtaining a video frame image, and extracting a brightness channel image, a blue channel image and a red channel image of the video frame image; taking the extracted brightness channel image of the associated video frame image as the input of an image super-resolution reconstruction model, and outputting a predicted high-resolution image of the associated video frame image by the model; and carrying out image merging on the high-resolution image associated with the video frame image and the extracted blue channel image and red channel image associated with the same video frame image, and then converting the merged image into a super-resolution image in an RGB color space. The image super-resolution reconstruction method provided by the invention is simple in principle and high in reconstruction speed, experiments show that the image super-resolution reconstruction method can be completely suitable for the image super-resolution reconstruction of the video stream, and the reconstructed super-resolution image is clear and good in visual sense.

Description

Technical field [0001] The present invention relates to the technical field of image reconstruction, and in particular to an image super-resolution reconstruction method and device for real-time video streams. Background technique [0002] Image super-resolution (SR) reconstruction aims to reconstruct a high-resolution image containing rich details by inputting one or more low-resolution images. It is important in the fields of monitoring equipment, satellite images, and medical imaging. Value. Through super-resolution reconstruction we can make single or multiple unclear images clear. [0003] At present, most research on image super-resolution reconstruction technology focuses on single-image super-resolution reconstruction (Signal Image Super Resolution, SISR), which uses the rich information contained in an image and the visual priors from sample images. , identify important visual cues to fill in image details. The traditional SISR method is based on empirical algori...

AI Technical Summary

Problems solved by technology

[0005] However, the traditional SISR method takes a long time to reconstruct images, and cannot meet the requirements for real-time image super-resolution reconstruction of video streams

Most of the existing image super-resolution reconstruction methods based on deep learning have problems such as blurred reconstructed images, large amount of noise, poor visual perception, etc., and the reconstruction process is complicated, and it is also difficult to meet the real-time image super-resolution reconstruction requirements of video streams.

Images