Video deblurring method based on space-time pyramid network and adversarial natural prior

Abstract

The invention discloses a video deblurring method based on a space-time pyramid network and adversarial natural prior. The method comprises the following steps: processing motion blurred video data collected in an actual environment, mapping the motion blurred video data to a Ycbcr color space, and independently inputting Y channel data into a space-time pyramid module to extract time domain and space domain information of different sizes of a blurred video; inputting the feature map extracted by the space-time pyramid module into a feature reconstruction module to process and reconstruct features to obtain deblurred video frames; performing calculation on y-channel deblurred video data obtained after space-time pyramid network processing of the previous space-time pyramid module and corresponding clear video data to obtain video natural prior information, and improviing the detail capture capability of the discriminator for video images. According to the technical scheme, after the fuzzy video is processed through the space-time pyramid network and the adversarial natural prior technology, a higher quantitative index and a better visual effect can be obtained.

Description

Technical field [0001] The invention relates to the technical field of video processing, in particular to a video deblurring method based on a spatiotemporal pyramid network and countering natural priors. Background technique [0002] With the rapid development of mobile Internet and electronic information technology, massive amounts of video data are generated every day, and video analysis and processing technologies are extremely valuable for security and surveillance. However, in the process of real-time video shooting, often due to factors such as the shaking of the shooting equipment and the motion of the shooting target, the captured video is blurred to a certain extent, which seriously affects people's acquisition of valuable detailed information from the video. But in many cases, the quality of the video content seriously affects the applicability of the video, for example: self-driving cars, road monitoring, security monitoring and other scenarios. [0003] Blurred video ...

AI Technical Summary

Problems solved by technology

[0004] With the deepening of research, motion video image deblurring technology has made some progress, however, there are still some key problems to be solved

Most of the current video image deblurring algorithms are aimed at space-invariant motion blur, and are not suitable for other blurring situations; the restoration algorithm will be affected by noise, and noise pollution will cause the deblurring algorithm to extract wrong information during the processing; finally , the current deblurring algorithm generally uses the deconvolution method to estimate the potential clear video image, which has a large amount of calculation and is poor for time-domain information acquisition.

Images