Video denoising method and device based on WT (Wavelet Transform) and block search

Abstract

The invention provides a video denoising method and device based on WT (Wavelet Transform) and block search; the method comprises the following steps: processing direction WT to video sequence to obtain the transformation coefficient; selecting the grain region and the smooth region in the video sequence according to the WT coefficient; processing double-sided contraction treatment to the WT coefficient; and obtaining the denoising video sequence based on direction wavelet by inverse transformation; denoising the grain region in the image domain based on block search; selecting the smooth region in the denoising result based on direction wavelet to overlap with the grain region obtained by denoising based on block search so as to obtain the final denoising video. According to the denoising method, the smooth region and the grain region in the video can be distinguished to respectively process through different denoising methods, thereby guaranteeing the accuracy of the grain region, guaranteeing the visual continuity between frames of the smooth region and obtaining obvious denoising effect.

Description

technical field [0001] The invention relates to the technical field of digital video denoising, in particular to a video denoising method and device based on wavelet transform and block search. Background technique [0002] With the development of computer and digital camera technology, video data affects people's life more and more widely. Post-processing has become a research hotspot in the field of video processing. [0003] Noise widely exists in images and videos, and there are many possible sources. For example, shot noise, thermal noise, and channel noise that exist in electronic hardware, and distortion and artifacts that occur when images and videos are compressed. The existence of noise not only reduces the visual quality of video, but also reduces the efficiency of video coding and the compressibility of media files for uncoded video. Traditional denoising algorithms generally include spatial filtering and temporal filtering. The advantages and disadvantages o...

AI Technical Summary

Problems solved by technology

The so-called non-local filtering refers to making full use of the redundancy of the natural scene itself in the denoising process, such as regular brick walls and repeated line information of the image outline. When these features are localized, non-local redundancy can be obtained and the filtering performance can be improved. Effect, then, the processing method of this block cannot be considered globally, and it will also cause the flickering problem of spatial filtering

Images