Classification of SAR Images Based on Pulse Coupled Neural Networks : Case on L-VH Band and C-VH Band
Di pos oleh deyani -
Volume
Abstrak
This paper described the result of classification on synthetic aperture radar images based on Pulse Coupled Neural Networks. Three textural feature, correlation, angular second moment and dissimilarity are processed by PCNN. The processed image in this research is image from East of Kalimantan. Each iteration by the PCNN gave one binary image which white pixels represents one type of the land covers. The contribution of this paper is the optimalisation on the variable of the equations of the PCNN and the integration of textural feature to the PCNN. The results of this experiment showed that classification based on PCNN and texture’s feature is better than classification based on PCNN. These two experiments extracted three types of land cover, forest, open area and city, water area consist of swamp, river and sea. The result of classification is matched to groundtruth.Tahun
2013
Penulis
Harwikarya
Keyword
Classification, PCNN, SAR, Texture