UC Berkeley Video and Image Processing Lab

Public Binary Code Release of Scalable Video for Internet Protocol (SVIP)

Most video compression methods used for real-time Internet transmissions are are neither bandwidth-scalable nor error-resilient. This produces a constant volume of inter-dependent packets that are prone to error propagation.

Producing a constant volume of traffic has several disadvantages. First, it would lead to congestive collapse when the aggregate bandwidth of the video traffic exceeds network capacity. Second, it competes unfairly with other adaptive traffic, such as TCP, that reduces transmission rate in face of network congestion. For shared environments like the Internet, it is important that users do not exceed their fair share of resources.

The use of non-error-resilient compression makes it necessary to employ error control mechanisms at the transport level. This typically takes the form of forward error correction (FEC) or retransmission. Retransmission based error control methods fail to be real-time, and effective use of FEC over the Internet requires a priori knowledge of the channel conditions.

An attractive alternative is to use scalable video compression with feedback rate control whereby transmission sources adjust their rates in response to changing network conditions. This is typically done by measuring packet loss rate or changes in round-trip delay. Such schemes are therefore reactive and do not prevent packet loss. As a result, error resilience remains an important issue.

SVIP is designed specifically to address the forementioned problems of flow and congestion control. Specifically, the compressed bit-stream under SVIP is packetized into packets with the following characteristics:

More detailed description as well as some experimental results are available in [1,2].

A free experimental software codec is provided by the Video and Image Processing (VIP) Laboratory in University of California at Berkeley. Click here to download the codec

The software was based on an earlier version of scalable video [3] written by David Taubman, and is subsequently refined by Dan Tan, Edward Chang, and Professor Avideh Zakhor. The binary code is compiled under Solaris 2.5 and the decoder requires Tcl/Tk 8.0 run-time libraries for displaying images. For more information and bug reports, please contact Professor Zakhor (avz@eecs.berkeley.edu) or Dan Tan (dtan@eecs.berkeley.edu).

[1] W.Tan and A.Zakhor. Internet Video using Error Resilient Scalable Compression and Cooperative Transport Protocol, Proc. ICIP, Vol. 3, pp 458-462, October 1998. [Compressed Postscript]
[2] W. Tan, E. Chang, and A. Zakhor. Real Time Software Implementation of Scalable Video Codec, Proc. ICIP, Vol. 1, pp 17-20, September 1996.[Compressed Postscript]
[3] D.Taubman and A.Zakhor. Multirate 3-D Subband Coding of Video. IEEE Trans. Image Proc., Vol. 3, No. 5, pp 572-88, September 1994.

UCB Video & Image Processing Lab / Nelson L. Chang / 16 September 1998
Copyright © 1998 by the Regents of the University of California