Digital lmage Processing for Meteorogical Satellites lnteract Aproach

This article shows the approach to the digital immage processing of TIROS - N/NOAA satellite data which is developed under the INTERACT project. The theme of the discussion is the problem of automatic oloud classification from the high resolution AVHRR data. The algorythm for cloud classification based on multivarite normal density as decision function is prorposed. Classifier learning phase is connected with use of training sample data from which the moments for normal density function are calculated. The decision area for the classification is defined by means of parameters (mean and covariances). The classification is performed using the Byes maximum likelyhood rule