Clouds Height Classification Using Texture Analysis of Meteosat Images

Abstract

In the present work, pattern recognition is carried out by the contrast and relative variance of clouds. The K-mean clustering process is then applied to classify the cloud type; also, texture analysis being adopted to extract the textural features and using them in cloud classification process. The test image used in the classification process is the Meteosat-7 image for the D3 region.The K-mean method is adopted as an unsupervised classification. This method depends on the initial chosen seeds of cluster. Since, the initial seeds are chosen randomly, the user supply a set of means, or cluster centers in the n-dimensional space.The K-mean cluster has been applied on two bands (IR2 band) and (water vapour band).The textural analysis is used where six parameters are calculated from the Co-occurrence matrix. These parameter were inserted in the K-mean. The best classifier feature is the angular second moment. When we use the angular second moment is used with any textural feature a good result were obtained for cloud classification, since the angular second moment gives indications on cloud homogeneity.