Real-Time Video Coding

Background

The objective of the Real-Time Video Coding group is to research and develop intelligent multimedia software components and delivery systems that adapt to congested and low infrastructure network environments. The emphasis is on jointly-optimised or co-operative behaviour between the delivery context, as in real-time network conditions, and the video/audio encoder rate control mechanisms. Innovative solutions are sought to enhance the Internet video experience in a developing world context where bandwidth is a scarce resource.

Current research within the group is concentrating on optimal and weighted context-based bit allocation techniques and implementations within international standard video codecs. A scalable and bandwidth adaptive multimedia broadcasting platform is under development to encompass the novelty of the bit allocation research.

Contact Details

Contact Person:	Dr Keith Ferguson
Tel:	+27(0)12 841 4433
Fax:	+27(0)12 841 4720
Email:	kferguson@csir.co.za

Seminars

30 September 2007 - IEEE students seminar at UCT

13 March 2008 - Seminar at UKZN

14 May 2008 - Guest lecture on Video Coding at University of Pretoria

Papers

Papers will be available here once complete.

Downloads

A part of the objective of the Real-Time Video Coding group is to utilise open source frameworks and to contribute to the open source community as well as to provide source code that is available for general usage. Please send comments, feedback, suggestions to rtvc[at]meraka.org.za.

DirectShow Filter Downloads

The following DirectShow filters are available for download as DLLs. These need to be registered using the regsvr32 command. Download the documentation below for more information about the filters.

Filter	Version	Description
Download Icon ScaleFilter	v1.1	The ScaleFilter is a DirectShow filter that can scale RGB24 media to specified target dimensions.
Download Icon CropFilter	v1.3	The CropFilter is a DirectShow filter that can crop RGB24 and RGB32 media.
Download Icon RGBtoYUV420Filter	v1.0	This filter converts RGB24 or RGB32 media to a YUV4:2:0 Planar format whilst retaining the original VideoInfoHeader of the media.
Download Icon YUV420toRGBFilter	v1.0	This filter converts the custom packed YUV420 planar format back to RGB24.

Source Downloads

• Base classes and interfaces - DirectShow base classes and interfaces have been written which simplify the creation of a transform filter. The CCustomBaseFilter class provides a base class for filters with a single input and output pin. The MultiIOBaseFilter project provides support for writing DirectShow transform filters with multiple input and output pins such as multiplexers and demultiplexers. All Meraka RTVC filters inherit these base classes, which are available in the download. The base classes have been compiled into a static library which the user can reference in the project.

• Filter Source Code - The download also contains the source code for the above-mentioned scaling, cropping and color conversion filters as well as the source code for an RTSP audio source filter which can play PCM audio streams.

• Image Utilities - The ImageUtils folder contains source code for various image processing such as cropping, scaling, etc. These files have been compiled into a static library which is also referenced by filters that require such functionality.

Download Icon Source Code v1.3

Download Icon Documentation

Links

• DirectShow related
  • Blog containing free DirectShow filters and GraphStudio, which offers some improved functionality over GraphEdit
  • DirectShow training

• RTP/RTCP/RTSP related
  • Open source RTP/RTCP/RTSP library that covers framing, sending of many media formats with a high level of importance based on standards. Main support via mailing list.

• Consortium Wiki (secure login required)
  • [1] This is the wiki for project information related to the Bandwidth Adaptive Real-Time Video Broadcasting over Internet project, known as ARTIST

Media

• CSIR researcher profile series: Streamlining delivery of internet video broadcasting
• DST-Innovation Fund investment boosts collaborative research on real-time video broadcasting project
• Innovation Fund investment boosts collaborative research on real-time video broadcasting project