Title：Video blind denoising method based on deep learning, computer device and computer-readable storage medium

Country：USA

Patent No.：US20210327031 A1

Legal Status：Published

Inventor：Xiang Xie, Shaofeng Zou, Guolin Li, Songping Mai, Zhihua Wang

Assignee：Tsinghua University

Address：

Filing Date：2021-10-21

Issue Date：

A video blind denoising method based on deep learning, a computer device and a computer-readable storage medium. The method includes: taking a video sequence from a video to be denoised, taking the middle frame in the video sequence as a noisy reference frame, performing an optical flow estimation on the image corresponding to the noisy reference frame and each other frame in the video sequence, to obtain optical flow fields; transforming, according to the optical flow fields, the image corresponding to each other frame in the video sequence to the noisy reference frame for registration respectively, to obtain multi-frame noisy registration images; taking the multi-frame noisy registration images as an input of a convolutional neural network, taking the noisy reference frame as the reference image, performing iterative training and denoising by using the noise2noise training principle, to obtain the denoised image. This solution may achieve the blind denoising of a video.

Patent Certificate： PDF/Jpg