Advanced computational methods for spatial information extraction

摘要: Abstract A variety of mathematical approaches for spatial information extraction using digitized aerial photography and satellite imagery have been developed implemented on serial computers. However, because data volume scale, the computational demands analysis procedures frequently exceed capacity available processing technologies. One way addressing this problem is through parallel in which power multiple computing units can be used a single problem. In study we investigate utility feature extraction. Our testing situation texture cooccurrence matrix indicates that dramatic reductions execution time are possible—an image required about 34 min to process one processor was solved under 2 nineteen processors. The availability additional processors could result smaller times. This speedup potential critical element future studies focusing more complex procedures.