Use of parallel processors in a photo based image generator
S. Barsamian · IEEE Conference on Aerospace and Electronics · 2002
An efficient real-time algorithm which maps three-dimensional features into photographic pixel coordinates is dealt with. The algorithm is optimized for use on parallel processors such as transputers. It efficiently processes polygonal terrain data based on DMA (Defense Mapping Agency) DTED (digital terrain elevation data). The same algorithm also maps cultural feature polygons, such as buildings or models. The algorithm's output structure feeds a unique hardware architecture which converts the output structures into world addresses at the pixel rate. Nonlinear mapping of polygon surfaces is included. Range information is also calculated to resolve occulting priorities. Transputers, with their excellent parallel processing capability, are well suited to process the algorithm. A transputer architecture allows near linear expansion of polygonal mapping capability as a function of the quantity of transputers used in the processing. Physical and virtual links between transputers are combined to create an efficient architecture. The algorithm and efficient hardware processing can produce a real-time photo-based image generator. The use of transputers in the implementation of an elevation processor is discussed in detail. The division of the elevation processor task into individual processes is considered.>