YUV Segmentator / Segmentation Tool

Easy to use tool for semiautomatic segmentation

Download:

Short Description

Video segmentation is a very important task in video filtering and video compression. The basic task of video segmentation is to break frames on layers correspondent to real objects and/or human perception of objects. It is very hard and can not be correctly solved automatically in general. Therefore a semiautomatic approach is commonly used now in real working solutions requiring good segmentation.

In a semiautomatic approach a user must specify an initial selection of objects of interest. Whereupon the selection is refined automatically. A user can correct selection and repeat its refinement.

We have developed an algorithm and a tool for semiautomatic video segmentation (YUV Segmentator). Using the tool, a user specifies initial selection in GUI with mouse. The selection is refined iteratively by the tool and tracked automatically to the following frames. A user can correct any segmented frame and restart the segmentation process from it.

Examples of Work

The first example shows result of result of ‘mother&daughter’ sequence segmentationusing the tool. The first frame is marked by a user and refined by the tool. The frame #150 segmentation is the output of the tracking algorithm.

The second example shows a sample result of ‘X-files’ movie segmentationusing the tool.

Primary Applications

  • Video cutting and editing
  • Special video effects
  • Regions of interest filtering
  • Intelligent video compression (using objects/layers)

Key Features

  • Automatic initial selection refinement
  • Several objects can be selected and tracked
  • Adjustable speed/quality trade-off

Basic Deliverables

  • Source code for a reference implementation in C and C++
  • C/C++ and assembly language sources of an implementation optimized for the PC (if required)
  • Algorithm description
  • Software description
  • Verification instructions

Contact Information

For information on licensing, please contact our sales staff.