Design and Development of a Coarse-Grained Parallel Desktop Computing System for Enhanced Image Processing

The use of imagery in military and civilian applications has expanded greatly in recent years. This growth has been driven primarily by the digital revolution in microelectronics, which has provided the signal processing, memory storage, and communications technology that make the wider use of images possible. The very ability to collect large numbers of images in near-real time creates another challenge - how to scan all the images and find objects of interest. The result is a need for high-speed digital signal and image processing systems and large mass memory storage devices to process, store, and analyze these data. To meet this need, ORINCON proposes during Phase I to complete the following tasks: (1) design of a coarse-grained parallel computer workstation architecture suitable for image processing and based on commercial off-the-shelf (COTS) hardware, (2) demonstration of possible 2-D and 3-D visualization techniques for transformed images, (3) design and documentation of an integrated Operator-Machine Interface (OMI), and (4) demonstration of a prototype for this image processing workstation. Phase II efforts will focus on the implementation of a 20-GFLOPS design capable of processing at least six different image types in near-real time.