A shape-adaptive partitioning method for MPEG-4 video encoding

MPEG-4 is a new standard for multimedia applications. Due to the flexible and extensible features of MPEG-4, the software-based implementation seems to be a natural and viable option. While such approaches usually require huge computing power, we can overcome such problem by using parallel and distributed processing. Because the behaviour of MPEG-4 objects may vary with time and such variation cannot be predicted in advance, the issues of data partition and load balancing of the multiprocessor systems need to be addressed carefully in order to achieve real-time operation performance. In this paper, we propose a shape-adaptive data partitioning method to guarantee the load balancing among the multiprocessor systems. The effectiveness of this method has been demonstrated by the experimental results.