JVT

ITU-T Q.6/SG16, also known as the Video Coding Experts Group (VCEG), while standardising three versions of the video compression standard H.263 during the 1990s, also worked on a long term video compression technology referred to at the time as H.26L. During the same time period, ISO/IEC JTC1/SC29/WG11, commonly known as the Moving Picture Experts Group (MPEG) developed two versions of IS 14496-2, more usually known as MPEG-4 video. In the summer of 2000, MPEG issued a call for proposals for new video compression technology, and evaluated the proposals with subjective tests one year later, finding VCEG?s H.26L algorithm the clear winner. This led to the decision to develop this technology jointly for the benefit of both organisations, and the creation of JVT, the joint video team of MPEG and VCEG.

 The terms of reference, http://www.itu.int/ITU-T/studygroups/com16/jvt/docs/JVTToR.pdf, were agreed, and the first meeting of the JVT was held in Pattaya, Thailand in December 2001. Work progressed rapidly and intensely in the JVT, with the first version being adopted in 2003, as Recommendation H.264 by the ITU-T, and as IS 14494-10, informally AVC (Advanced Video Coding) by ISO.

 This first version of the standard defined three profiles:

•  Baseline ? targeted at low delay and low computational complexity applications
• Extended ? targeted at low computational complexity but delay tolerant applications
• Main ? targeted at all other applications

 Work continued in the JVT, initially considering extensions to the standard for higher chrominance resolutions and higher bit depths than the 4:2:0 8 bit sampling already supported, but also to develop an extension of Main, known as the High profile, that achieved improved subjective picture quality on high resolution video. These extensions were adopted in 2005, and are known as FRExt, the Fidelity Range Extensions.

 Currently the JVT is developing scalable extensions of the standard, sometimes known as layered coding, where the concept is to encode a video bitstream into a hierarchy of ?layers?, where a receiver, starting with the lowest or ?Base? layer, can select to receive more layers, and hence more bits, to achieve better video quality. Each ?layer? can improve either temporal resolution (the number of pictures per second) and/or the spatial resolution (the quality of pixels and/or the number of pixels). The scalable extensions are due to be finished by the JVT at its April 2007 meeting, and formally approved by the ITU-T and ISO soon afterwards.

 JVT is also working on Multi-View Video Coding, which is concerned with the application scenario of a video scene being captured by many video cameras simultaneously, and these ?views? being compressed relative to each other into a single bitstream. The cameras would typically be displaced horizontally from each other, either on a straight line or an arc. The receiving device may decode and display one view, selected by the user, on a conventional display; or may decode and display multiple views on a multi-view display, thus giving different stereo views according to where the viewer is in relation to the display. The state of the art is for displays to be able to display about eight views simultaneously. This work is still at an early stage of development, and is due to be finished at the JVT?s April 2008 meeting.

 Participation in JVT meetings is open to, and contributions are accepted from, all parties qualified for participation in meetings of either MPEG or ITU-T. In addition, experts may be invited by the Chair, as permitted by ITU-T | ISO/IEC policies.

 JVT documents are available on the ftp site ftp://avguest:Avguest@ftp3.itu.org/jvt-site with username avguest password Avguest. They are also available by anonymous http at http://ftp3.itu.org/av-arch/jvt-site. The JVT terms of reference do allow for tighter control of input contributions, but to date no contributor has requested this.

 The JVT uses a number of e-mailing lists hosted by RWTH Aachen, the main one being jvt-experts@lists.rwth-aachen.de. These lists can be joined using the site https://mailman.rwth-aachen.de

 The planned schedule for future JVT meetings is as below.

 

Date	Location	Held under the auspices of:
21-27 April 2007	San Jose, USA	MPEG
2-6 July 2007	Geneva	ITU-T SG16
22-26 October 2007	Shenzhen, China	MPEG
14-18 January 2008	India?	MPEG
6-10 April 2008	Geneva?	MPEG & ITU-T SG16
21-25 July 2008	Hannover	MPEG
13-17 October 2008	Seoul? Korea	MPEG