RUSSIAN

MSU Deblur

MSU Graphics & Media Lab (Video Group)

Project: Dr. Dmitriy Vatolin
Implementation, ideas: Anatoliy Zapadinskiy
Konstantin Strelnikov

Often result video has blur artifact when scene with rapid movement is shot using video camera with large exposure time. Such artifact is named Motion Blur. The artifact apparents to the naked eye on plasma panels, CRT and LCD displays. Motion Blur spoils video image very hard. Typical example of the Motion Blur is shown on the next picture.

fast camera motion fast object motion
Motion Blur as result of fast camera motion Motion Blur as result of fast object motion

We have developed algorithm to remove Motion Blur artifact. Basic advantage of our Motion Deblur algorithm is automatic estimation of the Motion Blur parameters for each pixel in frame. True pixel value is reconstructed using estimated parameters. Algorithm sole parameter which is defined by user is exposure time.

Special attention is devoted to secondary contours processing and hard motion areas filtration. Secondary contours are typical problem of deblurring algorithms. Usually secondary contours become apparent nearby sharp edges. Error restoration in hard motion areas to reduce to pronounced artifacts.


Comparison


Image processing

Lucy-Richardson algorithm is used as competitor for our decision. Motion Blur parameters are defined by user.

source image processed by LR processed MSU Deblur
Source image with Motion Blur Processed by Lucy-Richardson Processed by MSU Deblur

Video processing

We don't known Motion Deblurring decisions for video processing. Therefore Lucy-Richardson algorithm with our Motion Blur parameters estimation algorithm without hard motion areas filtration is used as competitor for MSU Deblur. Basic Lucy-Richardson algorithm demands Motion Blur parameters definition by user.

source frame processed by LR processed MSU Deblur
Source frame with Motion Blur Processed by Lucy-Richardson Processed by MSU Deblur

source frame processed by LR processed MSU Deblur
Source frame with Motion Blur Processed by Lucy-Richardson Processed by MSU Deblur


Results


Some examples of MSU Deblur work.

source frame processed MSU Deblur
Source frame with Motion Blur Processed by MSU Deblur

source frame processed MSU Deblur
Source frame with Motion Blur Processed by MSU Deblur

Download


Please contact us if you are interested in a commercial license.

E-mail:  


Other resources


Video resources:

Last updated: 12-May-2022


Server size: 8069 files, 1215Mb (Server statistics)

Project updated by
Server Team and MSU Video Group

Project sponsored by YUVsoft Corp.

Project supported by MSU Graphics & Media Lab