Share Email Print
cover

Proceedings Paper

Software-based H.263 video encoder using a cluster of workstations
Author(s): Shahriar Akramullah; Ishfaq Ahmad; Ming Lei Liou
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

In this paper, we present a software based H.263 video encoder using a cluster of workstations. H.263 is an international standard for video coding using low bitrate communication. Due to a high computational cost, a real-time software based codec (specially encoder) is impossible to implement using currently available single-processor systems. Parallel processing using multiple computers, however, is a natural way of implementing a software-based real-time codec. A cluster of workstations is analogous to a parallel machine consisting of several processors each with its own memory and is a cost-effective computing platform. The workstations connected via an ATM switch use message passing interface (MPI) library for communication. Our implementation is based on a data-parallel approach in which parallelism is exploited within each frame of the video on a macroblock basis. Since the encoder is implemented in software it provides flexibility to enhance performance through modifications and improvements in various components. The encoder can be used with various numbers of processors connected via any physical hardware topology, and its peformance scales accordingly. It provides options for user-defined parameters such as the number of processors, the size of motion search window, quantization parameter, bit- rate, etc. The experimental results indicate that our approach is suitable for real-time applications such as video telephony on conventional analog telephone lines. An encoding rate of 30 frames/sec. (with QCIF format) is achieved using 12 workstations.

Paper Details

Date Published: 19 September 1997
PDF: 8 pages
Proc. SPIE 3166, Parallel and Distributed Methods for Image Processing, (19 September 1997); doi: 10.1117/12.279622
Show Author Affiliations
Shahriar Akramullah, Hong Kong Univ. of Science and Technology (United States)
Ishfaq Ahmad, Hong Kong Univ. of Science and Technology (Hong Kong)
Ming Lei Liou, Hong Kong Univ. of Science and Technology (Hong Kong)


Published in SPIE Proceedings Vol. 3166:
Parallel and Distributed Methods for Image Processing
Hongchi Shi; Patrick C. Coffield, Editor(s)

© SPIE. Terms of Use
Back to Top