Title :
Fuzzy min-max neural network for satellite infrared image clustering
Author :
Goswami, Bhaswati ; Bhandari, Gupinath ; Goswami, Suparna
Author_Institution :
Dept. of Comput. Applic., Narula Inst. of Technol., Kolkata, India
fDate :
Nov. 30 2012-Dec. 1 2012
Abstract :
The process of estimation of precipitation from satellite images begins with the detection and identification of convective clouds. Clustering of the satellite infrared images is required in order to estimate the cloud cover area. In this paper a neuro-fuzzy technique in the form of unsupervised fuzzy minmax clustering neural (FMMCN) network has been implemented for clustering satellite infrared image. Each cluster is in the form of an n-dimensional hyperbox defined by minimum and maximum points and a fuzzy membership function. FMMCN suits this application area because it is completely unsupervised and hence, unlabeled data can be used with it. Also the number of clusters is not required to be mentioned at the beginning as it is calculated dynamically.
Keywords :
clouds; fuzzy neural nets; fuzzy set theory; geophysical image processing; infrared imaging; minimax techniques; pattern clustering; cloud cover area; convective cloud detection; convective cloud identification; fuzzy membership function; fuzzy min-max neural network; n-dimensional hyperbox; neuro-fuzzy technique; precipitation estimation process; satellite infrared image clustering; unsupervised FMMCN; Clouds; Clustering algorithms; Image segmentation; Indexes; Labeling; Neural networks; Satellites; clustering; convective cloud; fuzzy cluster validity index; fuzzy min-max neural network; infra red images; k-means;
Conference_Titel :
Emerging Applications of Information Technology (EAIT), 2012 Third International Conference on
Conference_Location :
Kolkata
Print_ISBN :
978-1-4673-1828-0
DOI :
10.1109/EAIT.2012.6407913
