4 Chapter 1
some properties in common, we can then easily differentiate buildings of different
models. Similarly, we would not have any difficulty to identify alphanumeric
characters even when they are
of different fonts and with different orientation and
size. We can also differentiate men from women; differentiate people who came
from west hemisphere from those from east hemisphere; differentiate trucks from
cars even with different models. This is because the former ones and the latter
ones are defined as different pattern classes for the specific problem.
Pattern recognition is a process of categorizing any sample of measured or
observed data as a member of one
of the several classes or categories. Due to the
fact that pattern recognition is a basic attribute of human beings and other living
things, it has been taken for granted for long time. We are now expected to
discover the mechanism
of their recognition, simulate it, and put it into action
with the modern technology to benefit the human beings. This book is dedicated
to the design of a system to simulate the recognition of the human being, where
the acquisition of information through human sensory organs, processing of this
information and making decision through the brain are mainly involved. Pattern
recognition is a ramification
of artificial intelligence. It is an “interdisciplinary
subject.” This subject currently challenges scientists and engineers in various
disciplines. Electrical and computer scientists and engineers work on this;
psychologists, physiologists, biologists, neurophysiologists also work on this. A
lot of scientists apply this technology to solve problems in their own field, namely,
archaeology, art conservation, astronomy, aviation, chemistry, defenselspy
purposes, earth resource management, forensics and criminology, geology,
geography, medicine, meteorology, nondestructive testing, oceanography, surveil-
lance, etc. Psychologists, physiologists, biologists, and neurophysiologists devote
their effort toward exploring how living things perceive objects. Electrical and
computer scientists and engineers, as well as applied mathematicians, devote
themselves in the development of the theories and techniques for computer
implementation of a given recognition task.
When and where is the pattern recognition technique applicable? This
technique is usehl when (a) normal analysis fails; (b) modeling is inadequate;
and (c) simulation is ineffective. Under such situations, pattern recognition
technique will be found to be useful and would play an important role.
There are two types of items for recognition:
1. Recognition of concrete items. These types of items are visualized and
interpreted easier. Among the concrete items are spatial and temporal ones.
Examples
of spatial items are scenes, pictures, symbols (e.g., traffic symbols),
characters (e.g., alphanumeric, Saudi-Arabic character, Chinese characters, etc.),
target signatures, road maps, weather maps, speech waveform, ECG, EEG,
seismic wave, two-dimensional images, three-dimensional physical objects, etc.
Examples of temporal items are real time speech waveform, real time heart beat,