Graphics & Media Lab

Automatic Segmentation

MSU Graphics & Media Lab (Video Group)

Take a look at this article on the new site! Follow the link

Algorithm, ideas: Dr. Dmitriy Vatolin
Algorithm, implementation: Sergey Grishin,
Kostya Strelnikov, Maxim Makhinya, Sergey Putilin

Interest in advanced interactivity with multimedia data significantly increased last years. This cause an advent of new standards proposing the functionality for manipulation with multimedia data (an example of such a standard is MPEG4). That is why segmentation algorithms find its application in wide range of areas including content-based representation of multimedia data, improvement of coding efficiency in video compression standards, sophisticated query and retrieval of video and other content-based functionalities for multimedia applications.

Our developed algorithm performs detection and tracking of foreground (FG) objects in video. This is done by calculation of global motion with further estimation of local motion. Detection of a FG object position is then performed based on the information about global and local motion. The principal advantage of the method is its ability to detect a FG object even in case of ultra slow motion which is not common for algorithms of this type. Another important advantages include:

  • adjustable speed/quality trade-off
  • several segmentation precision levels
  • does not require manual segmentation


This section contains segmentation results of developed algorithm and its comparison with algorithm developed at University of Florida.

The first example (pic. 1, 2) demonstrates result obtained using 'dancer' test video sequence:

Original frame
Pic.1 Original frame
Segmentation result
Pic.2 Segmentation result

The second example (pic. 3, 4) shows result obtained using 'table tennis' test video sequence:

Original frame
Pic.3 Original frame
Segmentation result
Pic.4 Segmentation result

The next example (pic. 5, 6) shows segmentation result of 'bus' test video sequence:

Original frame
Pic.5 Original frame
Segmentation result
Pic.6 Segmentation result

Quality comparison of the developed method and algorithm of University of Florida is shown on the pictures below. This example shows results for test video sequence 'mother & daughter'. This sequence has two obstacles for successful segmentation. The first one is the proximity of colors belonging to different objects. And the second one (obstacle for foreground-background classification) is very slow motion of FG objects. Method of University of Florida produces segments consisting of parts actually belonging to several objects: the blue segment has parts in the area of woman's silhouette, blue segment points are presented around woman's head. However this comparison is not fully correct because algorithms perform segmentation of different types.

Original frame
Pic.7 Original frame
University of Florida result
Pic.8 University of Florida result
(different segments are marked by different colors)
Proposed method result
Pic.9 Proposed method result


For commercial license of this filter please contact us via

сontact email

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