High quality image and video resampling

Researchers: Alexey Lukin, Andrey Krylov, Anastasia Koulikova, Dmitriy Vatolin, Sergey Putilin

At the end of October 2002 we have started the project on “High quality image and video resampling”, supported by Samsung Advanced Institute of Technology (http://www.sait.samsung.co.kr/). The project’s goal was to design a new high-quality method for resampling of images and video.

The tasks of the project include:

Example of resampling of the image
Example of resampling of the image (down and up) using different algorithms (last two are our)

Our image resampling algorithm, G&M Lab Impress, has the following advantages over the standard methods:

Several metrics were implemented during the project (including LUV-metrics accounting for Contrast Sensitivity Function). One metric for measurement of image bluring-sharpening can be particularly interesting for future research.

Example of bluring-sharpening metric
Example of bluring-sharpening metric

Extensive testing of different image resampling algorithms has been performed. Seven “mira” test vector images were created for testing tasks. 9 photos provided by SAIT were also used. These 16 images have been processed using 18 resampling methods, (including methods of well known video-processing tool VirtualDub resampling methods and of Adobe Photoshop 7.0). Totally we have performed more than 1500 measurements for different methods, images and resolutions.

Different resampling methods metric example
Different resampling methods metric example

These images are obtained by comparison of scaled image with the original image (rasterized from vector image using Adobe Photoshop). Black color stands for no difference. Blue colors stand for slight differences. Green, yellow and red colors stand for more noticeable, significant differences.

Testing image for subpixel shift detection usage
Testing image for subpixel shift detection usage

E-mail: video@graphics.cs.msu.ru

Please read MSU filters FAQ before mailing.

10 Mar 2011
See Also
Learning-Based Image Compression Benchmark
The First extensive comparison of Learned Image Compression algorithms
MSU Video Group / Video data filtering and compession
VirtualDub MSU Logo Remover
The filter is intended to remove logo from films
MSU Stereo To Multiview 3D Video Conversion ( Glasses-free 3D-Displays Video Content creation)
Stereo to multiview video conversion algorithm for glasses-free autostereoscopic 3D displays
MSU Video Super-Resolution Filter
This filter increases resolution of video sequence while saving and improving details and reducing artifacts.
MSU Video Group / Video data filtering and compession
Site structure