| Summary: | Satellite-remote-sensing technologies have set off improvements in atmospheric research and developments of new tools in prospect of discovery to monitor. Classification of cloud image through satellite is well recognized as a valid approach in many climatic and environmental analyses. A multispectral cloud classifier was implemented to automate the interpretation of Kalpana-1 satellite image. In this paper, a novel image-clustering method, grounded on fuzzy statistics-based affinity propagation (FS-AP), has been proposed. It entails two steps: feature extraction and clustering. The objective is to study the volatility of the FS-AP for the classification of satellite cloud images optimally. Methods for classifying cloud type from satellite images are difficult in terms of efficiency and accuracy. Results show the effectiveness of the proposed technique for classification of satellite cloud image by comparing with fuzzy K-means and affinity propagation.
|
|---|
