Patlen (ohon appsoacho

) Template matching-

Template Matehing is used to determine the similarity between two entitics

(points, eurves, or shapes) of the same type. The
matehed with a stored pattern to be recognized is
template along with geometrical transformations. This
approaeh has some obvious disadvantages of being too rigid and having the need
for lots oftemplates
2) Statistical classification-

In this method, each pattern is represented in terms of some features or

measurements. The main
objective of this approach is to establish decision
boundaries in the feature
space. This separates patterns belonging to different
classes creating some rules for an
inter-class boundary.
3) Syntactic or structural matching-
This method works on a
hierarchy framework where, a pattern is said to be
composed of simple sub-paterns that are themselves built from yet simpler sub-
patterns. Considered equivalent to languages where primitives are
make words then lines than the alphabets which
page and then documents.
Types of Algorithms in Pattern
Recognition
Patten Recognition algorithms can be categorized as:
Statistical Techniques
Structural Techniques
Neural Network Approach
Template Matching
Fuzzy Model
Hybrid Model
Statistical Techniques
In Statistical modeling, modeling identifies where the
specific piece belongs. It is a
mathematical representation of observed data and refers to the use
learn from exàmples. Here the pattern is
of sttistics to
defined in terms of features and is selected
in a manner that different patterns take space without ovérlapping.
The selected features are crucial in the establishment of clusters. It
explains the
probability distribution, decision boundaries, etc., and then such patterns are
proposed to further processing. Then testing patterns are applied for the
recognition of patterns, leading to further classification methods.
Structural Algorithm Model
For complex pattern recognition, for instance, multi-dimensional entities, structural
algorithm models are best suited for. In this model, patterns are hierarchical in
nature, neaning they are categorized into subclasses. This model defines a
complex relationship between various elements. In this model, there are sub-
patterns, and the model is extended to structure and its forms in patterns. This
model is used for various purposes such as shape analysis, image analysis, and
others where measurable structures are contirmecd.
Neural Network Approach
Neural networks are used eficiently in classification. This approach employs
biological concepts to recognize patterns, and the result of this eftort is the
discovery of artificial neural networks.
For pattern recognition, the most common method is feed-forward networks, which
means that there is no feedback to the input. As humans learn from their past
experiences and mistakes, such networks also leam from their mistakes by giving
feedback to the input patterns. Such feedback assists in restoring the input pattems
and making them error-free. Here it is important to note that the neural networks'
performance improves as the hidden layers increase.

Template Matching
Template Matching involves the process of matching the target's features with the
predefined template and identifying the target by proxy. It is a technique used to
find a sub-image of a target image that resembles a template image.

This technique is úsed to define similarity among two similar objects. When the
patterm is matched, it is stored in templates, and the templates are given tlexibility
for scalar and rotational variations. The capability of this method depends upon the
already stored templates in the database.
Fuzzy Model
The fuzzy logic is a form of many-valued logic in which the truth values of
variables lie between 0 and 1, both inclusive. It is applied in order to handle partial
truth, where the truth value ranges between entirely true and entirely false. This
pattern recognition technique is known to produce the best end results conmpared to
other techniques because of the fact that here modeling is tor uncertain domains
and components for recognition.
llybrid Model
When multiple algorithms are
lybrid algorithm model. For deployed together, thattechnique
is known as thee
systems hidden Markov models
or instance, a combination
of neural nets with
learnng are all hybrid or a
combination of symbolic and fuzzy
such an
approach
strategies for pattern recognition. The best subsymbolic
is that it
is more employs the advantages of all the methods part of using
appropriate for hunman perception involved and
Pattern Recognition System
Pattern is
everything around in this digital world. A
physically or it can be observed pattern can either be seen
Recognition, pattern mathematically
In Pattern by applying algorithms.
is comprises
of the
things: following two fundamental
.Collection of observations
The
concept
Feature Vector:behind
the observation
The collection of
observations is also known as a feature
distinctive characteristic of a vector. A
items. Feature vector is good or service that sets it apart from feature is a
the similar
vector.The different classes combination of n features in
have different features n-dimensional column
always has the same featuresmay values.
values but the same class
Example:

Good Feature Linear Separability

Non-Linear Separability

tt
Bad Feature
Multi-Model Highly Correlated

(a) (b)
 D.fferentiate between good and bad
Feature propertics. features.

Classifier and Decision

In
Boundaries:
a
hypersurlace statistical-classification
a

that partitions the problem, a decision boundary is

A decision
boundary is the
underlying vector space into two sets.
output label of a classifier is region of a problem
space in which the
diserete-valued funetion that ambiguous.Classifier
is used to
is a
hypothesis or
to
particular data points. assign (categorical) class labels
2.
Classifier is used to
labeled decision partition the
feature space into clasS-
regions. While
between decision regions. Decision Boundaries are the borders

R1 R2 R1 R2
RA
R3

Classifier and decision boundaries

Components in Pattern Recognition System:

A pattern recognition systems can be
fivetypical components
partitioned into components.There are
for various pattern recognition systems. These
as following: are

A Sensor : A sensor is a device used to measure

as
pressure, position, temperature,
a
property, such
with feedback.
or
acceleration, and respond
A Preprocessing Mechanism : Segmentation is used and it is the
process of partitioning a data into multiple
be defined as the segments. It can also
technique dividing or partitioning an data into
of
parts called segments.
A Feature Extraction Mechanism:
feature extraction starts
from an initial set of measured data and builds
derived values
(features) intended to be informative and non-redundant,
facilitating the subsequent learning and generalization steps, and
in some cases leading to better human
manual or automated.
interpretations. It can be
 ADeseription Algorithm: Pattern recognition algorithms
generally aim to provide a reasonable answer for all possible
inputs and to perform "most likely" matching of the inputs, taking
into account their statistical variation
A Training Set : Training data is a certain percentage of an
Overall dataset along with testing set. As a rule, the better the
training data, the better the algorithm or classifier performs

FeedbackAdaplaton
Classifcation
Algorithm Class asslgnment

Feature
Real World- Sensor Preprocessing and Clustering >Cluster asslgnment
enhancement Extacbon Algorithm

Regresslon
Algoithm Predicted Varablefs)

APattern Recognition System

Design Principles of Pattern Recognition

In pattern recognition system, for recognizing the pattern or structure two
basic approaches are used which can be implemented in diferrent
techniques. These are -

Statistical Approach and

Structural Approach
Statistical Approach:
Statistical methods are mathematical formulas, models, and techniques that
are used in the statistical analysis of raw research data. The application of
statistical methods extracts information from research data and provides
different ways to assess the robustness of research outputs.
Two main statistical methods are used
Deseriptive Statistics: It summarizes data from a sample using indexes such as
the mean or standard deviation.
Inferential Statisties: It draw conclusions from data that are subject to random
variation.
Structural Approach:
The Structural Approach is a technique wherein the leamer masters the patlern of
sentence. Structures are the different arrangements of words in one accepted style
or the other.
Types of structures:
Sentence Patterns
. Phrase Patterns
Formulas
Idioms

Difference Between Statistical Approach and Structural Approach:

Sr. No. Statistical Approach Structural Approach

Statistical decision theory. Human perception and cognition.

2 Quantitative features. Morphological primitives

3 Fixed number of features. Variable number of primitives.

Ignores feature relationships. Captures primitives relationships.

Semantics from feature position. Semantics from primitives encoding.

5

6 Statistical classifiers. Syntactic grammars.

Decision region:
A decision region is an area or volume, marked by cuts in the pattern space. All
of the patterns within a usable decision region belong to the same class. As a
the location of a pattern identifying what decision region it lies in can be
-

result,
-

used to classify it.

The goal of a classifier is to partition feature space into class-labeled decision
regions. Borders between decision regions are called decision boundaries

Metric space
Metric space is a set together with a notion of distance between its elements,
usually called points. The distance is measured by a function called
a metric or distance function.

Distance measure

The distance measure is one of the proximity measures to classify an unknown

patternto its nearest class. In other words, it finds the dissimilarity between
patterns.