REFERENTIAL CLASSIFICATION - AN INTELLIGENCE BASED ALGORITHM
Othman Alhusain
Research Fellow, Department of Photogrammetry, Technical University of Budapest
Muegyetem rkp.3.1.24, H-1111 Budapest, Hungary
ISPRS Commission II, Working Group 2.
KEY WORDS: Image, Classification, Intelligence, Algorithm,
ABSTRACT:
This paper describes a recently developed method which can be used in the classification of pixels in satellite images.
The new method named "Referential Classification" is a concept in which the spectral contents of the pixels are not
considered absolutely; but instead they are correlated with a kind of ancillary data that involves the coordinations of
a corresponding pixels and their possible variations of spectral data base. The referential classification process is
carried out in two subsequent stages. The Application of this method in classifying remotely-sensed images resulted
in higher accuracy and less demand for the user intervention during the classification course which imply higher degree
of automating the classification process without sacrificing its accuracy.
KURZFASSUNG:
Im Artikel es geht um die neueste Methode, was im Satellitenbildklassifikation anwendbar ist. Die als "Referential
Classification" genannte Methode ist ein Konzept, bei welchem der spektrale Inhalt der Pixel nicht absolutweise
berücksichtigt wird, sondern statt dessen er mit verschiedenen Daten korreliert ist, die Koordinaten der entsprechenden
Pixel und ihre Variationen des spektralen Datenbasis enthält. Der Referential Classification-Prozeß wird in zwei
darauffolgenden Schritten durchgeführt. Die Anwendung der Methode in Klassifizierung von Fernerkundungsdaten
ergibt höhere Genauigkeit und braucht wenigeren Eingriff während der Klassifizierung, was ohne Genauigkeitsverlust
eine höhere Automatisationsstufe des Klassifikationsprozeßes bedeutet.
1. INTRODUCTION Second, is the ability of the human beings to draw and
assemble data making considerable use of previous
Classification of remotely sensed images collected by training and experience to make critical decisions (Hunt,
multispectral scanners onboard different kinds of 1984).
satellites has been performed by conventional methods Referential classification is a concept in which the
depending mainly on the spectral content of the pixels spectral contents of the pixels are not considered
while ignoring other kinds of information related to the absolutely; but instead they are correlated with the
pixels, depending only on one factor which is the coordinations of the corresponding pixels and their
spectral content of the pixel and discarding the role of possible variations of spectral database so enhancing the
other information about the pixel is a serious pitfall in probability of having correct classification of image
the classification procedure. However, there is a clear pixels (Alhusain, 1992).
evidence that if one like to have an efficient
classification then the classification concept must
simulate the human being high ability of perception and 2. REFERENTIAL CLASSIFICATION
analyzing (Baxes, 1985).
The role of the human beings in the interpretation of The algorithm implemented within this method first
image data is of unique and great value for two reasons: proceeds towards assigning Cartesian coordinates (x)
First, is the human being's ability to efficiently acquire to each pixel, figure 1 shows the flowchart of the
and maintain an awareness of a universal nature; an referential classification method. This might be done
awareness not only of the complete remote sensing data locally where the coordinates of each pixel are valid
but also of the associated reference data and the only within the image where the assignation process has
relationship between the reference and spectral data. been defined, or globally where the coordinates of each
332
pixel are
coordinat:
all previc
about eac
image. TI
design is :
about pix
change in
on pixels
but also
assignatio
of the imz
an image
PIX
Rep
reje
Li
| N
tn
O
HS
Figure -