Video denoising method and calculation equipment

Abstract

The invention discloses a video denoising method. The method includes the steps of obtaining at least one adjacent frame of a frame to be denoised in a video sequence; performing time domain denoising on each pixel point of the frame to be denoised, the time domain denoising including the steps of, for each adjacent frame of the frame to be denoised, obtaining matched pixel points of the pixel points of the frame to be denoised on the adjacent frame, and calculating the weights of the matched pixel points; and based on the weights of the obtained matched pixel points of the pixel points of the frame to be denoised on each adjacent frame, calculating the weighted average of the pixel points of the frame to be denoised and the matched pixel points thereof, so as to obtain pixel values of the pixel points of the frame to be denoised after time domain denoising. The invention also discloses corresponding calculation equipment and a computer storage medium.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a video denoising method and computing equipment. Background technique [0002] In the application scenario of image or video acquisition, the acquired image will inevitably have signal noise. For example, pictures or videos taken in scenes such as night usually have relatively obvious noise. Image denoising is a very important link in image processing (single image processing or video frame sequence processing). The main problem of denoising is how to effectively remove the temporal domain and spatial domain noise while effectively preserving the boundary texture details of the image from being destroyed. [0003] At present, the mainstream processing method is to perform bilateral filtering in the spatial domain and mean filtering in the time domain. However, the disadvantage of this method is that it cannot guarantee the consistency of the filtering strength of adja...

AI Technical Summary

Problems solved by technology

[0003] At present, the mainstream processing method is to perform bilateral filtering in the spatial domain and mean filtering in the time domain. However, the disadvantage of this method is that it cannot guarantee the consistency of the filtering strength of adjacent frames of the video. The intensity of the filter is inconsistent and flickering occurs

Among them, even if the average filter is used, it cannot solve the flicker noise in the time domain well. If there is motion in the two frames before and after the video, this average filter in the time domain will also cause strong ghosting.

[0004] In particular, for high-noise images, although choosing a larger smoothness can denoise better, it will lead to blurred images and their edges are relatively blurred

If you choose a small smooth, it is difficult to effectively remove the noise in the image

Images