Achievement

Consistent Map-albums from Remotely Sensed Imagery:

Consistent (multi-scale) map-albums are compatible collections of maps, each representing features of a common area with a different proportionate size. These are important for consistency of decisions made across levels of hierarchal organizations. Challenges arise from significant computation costs to evaluate candidate map models at different locations and scales using statistical maximum likelihood method. Our multidisciplinary team with geographers, statisticians and computer scientists designed a novel framework to reduce computational overhead using information about context. Evaluations with land-use classification at multiple spatial resolutions from satellite imagery show that the proposed approach reduces the computational cost significantly. Preliminary results were published in IEEE Workshop on Spatio-temporal Data Mining (2006) and complete results were accepted by Knowledge and Information Systems journal in 2009.