Multispectral Satellite Images Processing for Forests and Wetland Regions Monitoring Using Parallel Mpi Implementation

Abstract

The effective methods and algorithms based on fuzzy clustering for processing multispectral satellite images have been developed. To enforce discrimination of different land covers and to improve area separation additional channels with fractal characteristics have been evaluated and included into aggregate multichannel image. To significantly reduce time expenses parallel computing technique was used for practical implementation. The classification based on radial basis function (RBF) neural network has been presented. Data preprocessing techniques such as histogram processing and texture features calculation are discussed. The original approximation method based on radial basis functions was developed to create superimposing transform of multispectral images into/from geographical projections. All the software has been implemented as GIS GRASS modules to be runnable in Massive Parallel Processing (MPP) cluster environment using Message Passing Interface (MPI). Experimental testing of developed algorithms and techniques has been carried out using images received from Landsat 7 ETM+ Satellite.

3 Figures and Tables

Cite this paper

@inproceedings{Sadykhov2007MultispectralSI, title={Multispectral Satellite Images Processing for Forests and Wetland Regions Monitoring Using Parallel Mpi Implementation}, author={Rauf Kh. Sadykhov and Andrey V. Dorogush and Yegor V. Pushkin and Leonid P. Podenok and Valentin Ganchenko}, year={2007} }