The hot spots' which have the values unlike ones in the neighborhood often appear in the spatio-temporal data set in the field of space and planetary sciences. Method of automatic extraction of such hotspots as objects from the spatio-temporal data by using mixture distribution of multivariate normal distributions are described, which aims to understand changing patterns such as co-occurrence or temporal rules among the hot spots and to predict them. This paper also introduces the example system in which hotspots are extracted from meteorological satellite imagery. Plans for future work are also discussed based on the problem that became clear through this study example.
内容記述
形態: カラー図版あり
内容記述（英）
Physical characteristics: Original contains color illustrations
時空間変動データからのホットスポット自動抽出・要約システムの開発
AA1730023004.pdf
(2.92MB)
