| Resumen |
Up-to-date, within the Remote Sensing field,
many areas are oriented and specialized to particular
cases study, which are related to digital geo-image
processing. Unfortunately, the most of recent methods,
such as wavelets, fractal decomposition and analysis of
textures, loose the essential components of the image.
In other words, these approaches do not explicitly
represent the semantic structure of the images.
In this work, we point out an object oriented framework
that allows us to make a decomposition of layers, which
are generated from a remote source. Thus, our approach
preserves the semantic properties of the objects that
compose the source image.
Our approach is divided up two main stages: analysis
and synthesis. In the analysis stage, given any digital
image, generate the hierarchical representation of the
objects that compose the geo-image to process. This
stage is indispensable in the method, because it is
necessary to identify the segments of the geo-image
that are semantically important and considerable.
Later, a synthesis stage is required; because the output
of information from the analysis stage is described as a
set of pixels and it is isolated according to the
semantics, which is generated in this stage.
Nevertheless, it is most adequate to count on the same
information represented in an alternative form; thereby
it is possible to achieve the information compression.
Keywords: remote sensing, object-oriented analysis,
semantic dec |
