Automatic cloud classification of satellite imagery can be ofgreat help to meteorological studies. A neural network-based cloudclassification system is developed and introduced. Several imagetransformation schemes such as wavelet transform (WT) and singular valuedecomposition (SVD) are used to extract the salient textural feature ofthe data and is then compared with those of the well-known gray-levelco-occurrence matrix (GLCM) approach. Two different neural networkparadigms namely the probability neural network (PNN) and theunsupervised Kohonen (1990) self-organized feature map (SOM) are chosenand examined. The performance of the proposed cloud classificationsystem is benchmarked on the Geostationary Operational EnvironmentalSatellite (GOES) 8 data set and promising results have been achieved
展开▼
