The following image has been blurred (motion blur) and noise has been added. The right picture shows the extraction result.
Among other images, we have tested SIOX against a benchmark that has been provided by
We added trimaps to better fit the inputs that a typical user would provide when using SIOX.
Click here if you are interested in them. The current overall error we calculated
using Microsoft's metric is: 3.59%. Please refer to our scientific publications below for
details on the bechmark as well as the algorithm.
We explictly welcome applications that integrate SIOX as long as they keep to the rules of the underlying open source licenses. If you want to take a look at source code there are three possibilities:
If you have any questions regardings the integration of SIOX into your application or research project, please contact us for further assistance.
- The SIOX Java Reference Implementation is aimed to provide the conceptual source code reference (slow, but easy to understand).
Download the SIOX Java Reference Implementation v1.60 (zip file, 250kB).
The SDK is released under the Apache License and works with Java 1.4 and higher.
The SIOX Java Reference Implementation contains SIOX in its very original form. All other implementations have to be considered derivates and might contain bugs, lack features, or have a modified functionality.
- If you want to see the current C sources (GNU GPL) from the GIMP go directly to the GIMP CVS repository.
- The current Inkscape implementation can be downloaded from their Sourceforge SVN.
For further details on the algorithm and benchmark results please download one of the following papers.
Most Comprehensive Work:
- G. Friedland: Adaptive Audio and Video Processsing for Electronic Chalkboard Lectures, PhD thesis, Department of Computer Science, Freie Universitaet Berlin, October 2006.
Still Image Approach:
- G. Friedland, K. Jantz, T. Lenz, F. Wiesel, R. Rojas: Object Cut and Paste in Images and Videos, International Journal of Semantic Computing Vol 1, No 2, pp. 221-247, World Scientific, USA, June 2007.
- G. Friedland, K. Jantz, L. Knipping, R. Rojas: Image Segmentation by Uniform Color Clustering -- Approach and Benchmark Results,
Technical Report B-05-07, Department of Computer Science, Freie Universitaet Berlin, June 2005 (PDF, 18MB).
- G. Friedland, K. Jantz, R. Rojas: SIOX: Simple Interactive Object Extraction in Still Images, Proceedings of the IEEE International Symposium on Multimedia (ISM2005), pp. 253-259, Irvine (California), December, 2005. Download PDF from IEEE Computer Society Digital Library.
- G. Friedland, K. Jantz, T. Lenz, R. Rojas: Extending the SIOX Algorithm: Alternative Clustering Methods, Sub-pixel Accurate Object Extraction from Still Images, and Generic Video Segmentation, Technical Report B-06-06, Department of Computer Science, Freie Universitaet Berlin, January 2006 (PDF, 10MB).
- G. Friedland, K. Jantz, T. Lenz, F. Wiesel, R. Rojas: A Practical Approach to Boundary-Accurate Multi-Object Extraction from Still Images and Videos, to appear in Proceedings of the IEEE International Symposium on Multimedia (ISM2006), San Diego (California), December, 2006.
- G. Friedland, R. Rojas: Anthropocentric Video Segmentation for Lecture Webcasts, EURASIP Journal on Image and Video Processing, Volume 2008, Hindawi Publishing Corporation, 2008.
- G. Friedland, K. Jantz, R. Rojas: Cut & Paste: Merging the Video with the Whiteboard Stream for Remote Lectures, Technical Report B-05-19, Department of Computer Science, Freie Universitaet Berlin, May 2005 (PDF, 4MB).
- N. Santrac, G. Friedland, R. Rojas: High Resolution Segmentation with a Time-of-Flight 3D-Camera using the Example of a Lecture Scene, Technical Report B-06-09, Department of Computer Science, Freie Universitaet Berlin, September 2006 (PDF, 9 MB).
- G. Friedland, R. Rojas: Human-Centered Webcasting of Interactive-Whiteboard Lectures, to appear in Proceedings of IEEE International Workshop in Multimedia Technologies for E-Learning, San Diego, California, December 2006.
Copyright 2005, 2006, 2009 by SIOX Team. All rights reserved.