Self-organizing GIS for solving problems of ecology and
landscape studying
Self-organizing GIS for solving problems of ecology and
landscape studyingNikolay G. Markov, Alexandr A. Napryushkin
Tomsk Polytechnical University,
GIS laboratory, Tomsk, Russia
e-mail: [email protected]
Self-organizing vector-raster GIS (SOVR GIS) solves the following tasks:
Preliminary processing of the received remote sensing (RS) data (solving tasks of projection transforming, geo-referencing, linear and nonlinear filtration, spectral and geometrical transformation)
Thematic processing of the processed RS data (automated interpretation)
Spatial analysis of the extracted thematic information represented in a vector format (complex quantitative estimations of the researched objects and phenomena)
Subsystem of preliminary processing
Subsystem of self-organizing
Subsystem of vector data visualization
Interface shell of SOVR GIS
Subsystem of interpretation and
vectorization
Subsystem of spatial analisys
Subsystem of raster data visualization
Data input-output
subsystem
Subsystem of 3D visualization
Raster component Vector component
Fig. 1. General structure of SOVR GIS
Thematic processing - the stage of extracting the geometric
information from preliminary processed aerospace images.
SOVR GIS provides the facilities for automatized extraction of thematic information
from aerospace images.
Fig. 2. Automatized extraction of thematic information from aerospace images by means of SOVR GIS
Kohonen’s neuronet classifier
Preliminary processed aerospace image
Vectorizing procedure
Recognition procedure
Self-organizing procedure
Textural analysis
procedure
Cartographic sources
Extended feature space
Training data
Vector thematic layers
Spatial analysis
Decisions
Self-organizing procedure
(decision making algorithm)
Non-parametric classifiers
Advanced Bayesian classifier
Extended feature space aerospace image
Recognized landscape
objects
Fig.3 Self-organizing procedure
Fig. 6. Mapping forest types of Tomsk region with SOVR GIS (satellite RESURS-0, MSU-E scanner)
Initial aerospace image Map
Cedar Pine tree Cedar+Fir
Classified image
Self-organizing GIS for solving problems of ecology and
landscape studying
Self-organizing GIS for solving problems of ecology and
landscape studyingNikolay G. Markov, Alexandr A. Napryushkin
Tomsk Polytechnical University,
GIS laboratory, Tomsk, Russia
e-mail: [email protected]