Parallel algorithms for image and video mosaic based applications
MetadataShow full item record
Image and video mosaic based applications have received growing interest in recent times. Image and video mosaicking techniques have been applied to various fields such as video compression, video enhancement, digital libraries, interactive video analysis, virtual environments, low-bitrate video transmission, and interactive video editing and manipulation systems. Despite all the improvements that have been made, image and video mosaicking is still a computationally intensive process. This characteristic is crucial when one considers deploying the mosaicking technique in a real time environment such as live video transmission. Two parallel programming paradigms for a video mosaicking algorithm are discussed in this thesis, namely, shared memory programming algorithm (SMPA) using POSIX threads and distributed memory programming algorithm (DMPA) using MPI. Experimental results from a Sun Fire server and a cluster of Sun Blade workstations show that our SMPA achieves a linear speedup in most cases and a super linear speedup on some occasions.