A Parallel Processing Framework for Real-Time Software-Only Video Compression for Broadcast and Mobile Services

作     者：Zheng, Lizhi Cosmas, John 

作者机构：Brunel Univ Sch Engn & Design Uxbridge UB8 3PH Middx England 

出 版 物：《JOURNAL OF INTELLIGENT SYSTEMS》 (智能系统杂志)

年 卷 期：2006年第15卷第1-4期

页      面：261-297页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：European Union under the Information Society Technologies program as the project CISMUNDUS (Convergence of IP-based Services for Mobile Users and Networks in DVB-T and UMTS Systems) 

主　　题：broadcast and mobile service parallel processing framework real-time video compression 

摘      要：This paper develops a generic parallel processing framework on embedded multicomputer systems for real-time video compression for convergent broadcast and cellular services. This framework allows easy prototyping of new parallel video compression algorithms and integration with video applications. Using the proposed parallel processing framework, a parallel H. 263 encoder has been implemented on a PowerPC processor-based Mercury RACE multicomputer. The experimental results show promising parallel speedups. Real-time performance was achieved for QCIF video using the non-optimized reference software. The parallel processing framework not only can be configured with different number of processors on an embedded multicomputer system but also is portable onto varied embedded multicomputer platforms. Therefore, this paper provides a software implementation solution as an alternative to hardware video compression for broadcast and mobile services.