Chapter 20
Pattern Recognition and Classification of
Remotely Sensed Images by Artificial Neural
Networks
G.M.Foody
20.1
Introduction
Pattern recognition is concerned with a range of information processing issues
associated with the description or classification of measurements. It is based on a
broad and often loosely related body of literature and techniques (Schalkoff
1992). Although statistical and structural (syntactic) approaches have dominated
the subject there has been a growing interest in the use of neural networks for
pattern recognition applications (Schalkoff 1992; Bishop 1995). This is
particularly evident in relation to pattern recognition applications in remote
sensing. Remote sensing is often used to derive information on the environment
(Campbell 1996; LilIesand and Kiefer 2000). For example, satellite remote
sensors are commonly used to provide images of the Earth's surface that may be
analysed to yield information on a diverse range of issues of ecological
significance. This includes information on a variety of important environmental
phenomena including land cover and its dynamics (e.g. deforestation), vegetation
productivity and yield, soil water content, water quality and variation in surface
temperature. Remote sensing has, therefore, the potential to provide information
to support ecological research, particularly that addressing macro or coarse spatial
scale issues (Roughgarden et al. 1991; Kasischke et al. 1997; Lucas and Curran
1999; Hall 2000). The value of remote sensing as a source of information on the
environment is, however, sometimes limited by the image analysis techniques
used. Traditionally, statistical techniques have been used widely in the analysis of
remotely sensed data. However, in common with other fields of study, including
ecology, the last 10-15 years has witnessed a rapid growth in the use of neural
networks (Atkinson and Tatnall 1997; Lek et al. 2000). This chapter aims to report
on some of the main application of neural networks in remote sensing and indicate
topics where further development may be expected to occur.
Pattern Recognition and Classification of
Remotely Sensed Images by Artificial Neural
Networks
G.M.Foody
20.1
Introduction
Pattern recognition is concerned with a range of information processing issues
associated with the description or classification of measurements. It is based on a
broad and often loosely related body of literature and techniques (Schalkoff
1992). Although statistical and structural (syntactic) approaches have dominated
the subject there has been a growing interest in the use of neural networks for
pattern recognition applications (Schalkoff 1992; Bishop 1995). This is
particularly evident in relation to pattern recognition applications in remote
sensing. Remote sensing is often used to derive information on the environment
(Campbell 1996; LilIesand and Kiefer 2000). For example, satellite remote
sensors are commonly used to provide images of the Earth's surface that may be
analysed to yield information on a diverse range of issues of ecological
significance. This includes information on a variety of important environmental
phenomena including land cover and its dynamics (e.g. deforestation), vegetation
productivity and yield, soil water content, water quality and variation in surface
temperature. Remote sensing has, therefore, the potential to provide information
to support ecological research, particularly that addressing macro or coarse spatial
scale issues (Roughgarden et al. 1991; Kasischke et al. 1997; Lucas and Curran
1999; Hall 2000). The value of remote sensing as a source of information on the
environment is, however, sometimes limited by the image analysis techniques
used. Traditionally, statistical techniques have been used widely in the analysis of
remotely sensed data. However, in common with other fields of study, including
ecology, the last 10-15 years has witnessed a rapid growth in the use of neural
networks (Atkinson and Tatnall 1997; Lek et al. 2000). This chapter aims to report
on some of the main application of neural networks in remote sensing and indicate
topics where further development may be expected to occur.