Vision Computation on Gigabyte Datasets Using Clusters. An investigation into the issues surrounding the application of clusters to the analysis of large datasets in the domain of computer vision.

As researchers approach a better understanding of the interdependence of multiple ecosystems, technologies capable of producing large and complex datasets will become increasingly important. New sensors will make large amounts of data available at little or no cost. Inexpensive disk and tape systems will allow anyone to store the data locally. However the scope of research may still be limited by the lack of affordable systems capable of processing all the available data. To make use of this wealth of data, researchers will have to find new ways of processing large amounts of data in a reasonable amount of time. We propose the design and development of a new system that will address the problems faced by researchers working with large datasets. This system will provide earth scientists and other researchers who have large datasets the ability to make use of task and data level parallelism with minimal setup overhead. This will reduce the processing time and allow them to process datase..