الفهرس 
Introduction to Image Understanding Systems.Only 14 pages are availabe for public view
 |  | from | 158 |  |  |
| | | from | 158 | | |
Abstract
This thesis deals with the problem of scene matching.
Given a pictorial description of a region of a scene, it is
desired to determine which region in another image is
similar. The simplest method to solve this problem, which is
called template matching, is described as it is the backbone
of all other methods.
A review of other image matching techniques using image
processing based algorithms such as ordered search techniques
in template matching, two-stage template matching, coarsefine
template matching, and sequential similarity detection
algorithms are discussed.
The most effecient algorithms for scene matching are
analyzed and discussed. Those are the sequential
hierarchical scene matching algorithms. The main target of
this thesis is to investigate those approaches. The first
approach is the basic sequential hierarchical scene matching
dealing with gray-scale images. The sequential decision
rules are discussed as well as the derivation of threshold
sequence. The” pairing functions” concept applied to the
second approach, namely, the sequential scene matching using
edge features, is described as well as edge extractions and
similarity measure.
Experimental results are presented for matching sattelite
images of AI-Minea (Egypt) and Montana (USA) using those
sequential hierarchical scene matching algorithms as well as
the two-stage template matching algorithm. The experimental
work is done using the Remote Image Processing System (RIPS).
The results prove eff1cency and success in reaching the best
match location with minimum required computations. A comment
on the results 1s presented.