This document discusses pattern recognition. It defines a pattern as a set of measurements describing a physical object and a pattern class as a set of patterns sharing common attributes. Pattern recognition involves relating perceived patterns to previously perceived patterns to classify them. The goals are to put patterns into categories and learn to distinguish patterns of interest. Examples of pattern recognition applications include optical character recognition, biometrics, medical diagnosis, and military target recognition. Common approaches to pattern recognition are statistical, neural networks, and structural. The process involves data acquisition, pre-processing, feature extraction, classification, and post-processing. An example of classifying fish into salmon and sea bass is provided.

1. PATTERN
RECOGNITION
1

2.  A pattern is an abstract object, or a set of measurements
describing a physical object.
 Pattern is defined as composite of features those are
characteristic of an individual.
2

3.  A pattern class (or category) is a set of patterns sharing
common attributes.
 A collection of “similar” (not necessarily identical) objects.
 During recognition given objects are assigned to
prescribed classes.
3

4.  Theory, Algorithms, Systems to put Patterns into
Categories
 Relate Perceived Pattern to Previously Perceived Patterns
 Learn to distinguish patterns of interest from their
background
4

5. HUMAN PERCEPTION
 Humans have highly sophisticated skills for sensing
their environment and taking actions according to what
they observe, e.g.,
 Recognizing a face.
 Reading handwriting.
 Understanding spoken words.
 Distinguishing fresh food from its smell.
And so…….
 We can give these similar capabilities to machines.
5

6. EXAMPLES OF APPLICATIONS
•Handwritten
•Printed texts
Optical
Character
Recognition
(OCR)
•Face recognition
•Finger prints recognition
Biometrics
•Medical diagnosis
•X-Ray, EKG etc.
Diagnostic
systems
•Automated Target Recognition (ATR).
•Image segmentation and analysis (recognition
from aerial or satelite photographs).
Military
applications
6

7.  Statistical way to Pattern Recognition: based on underlying
statistical model of patterns and pattern classes.
 Neural networks: classifier is represented as a network of cells
modeling neurons of the human brain (connectionist approach).
 Structural (or syntactic) Pattern Recognition: pattern classes
represented by means of formal structures as grammars,
automata, strings, etc.
7

8.  Grid by Grid Comparison
A A B
8

9. 9
A A B
Grid by Grid
Comparison
Grid by Grid Comparison

10. A A B
10
0 0 1 0
0 1 1 0
0 1 1 1
1 1 1 1
1 0 0 1
0 1 1 0
0 1 1 0
1 1 1 0
1 1 1 1
1 0 0 1
No of
Mismatch= 1
10
Grid by Grid Comparison

11. A A B
Grid by Grid
Comparison
11
Grid by Grid Comparison
11

12. A A B
12
0 0 1 0
0 1 1 0
0 1 1 1
1 1 1 1
1 0 0 1
1 1 1 1
1 0 1 1
1 1 1 0
1 0 1 1
1 1 1 1
No of
Mismatch= 5
Grid by Grid Comparison
12

13. PROBLEM WITH GRID BY GRID COMPARISON
 Time to recognize a pattern - Proportional to the number
of stored patterns ( Too costly with the increase of number
of patterns stored )
13
A-Z a-z 0-9
*/-+1@#
Solution Artificial Intelligence

14. ARTIFICIAL INTELLIGENCE
Two Phases :
 Learning and
 Detection.
14

15.  Supervised learning (Classification): a person or programmer
provides a category label or cost for each pattern in the training
set.
 Unsupervised learning (Clustering): the system forms clusters
or natural groupings of the input patterns.
15
Classification
(Supervised Classification)
Clustering
(Unsupervised Classification)

16. PATTERN RECOGNITION PROCESS
16

17. PATTERN RECOGNITION PROCESS
 Data acquisition and sensing:
 Measurements of physical variables.
 Important issues: bandwidth, resolution , etc.
 Pre-processing:
 Removal of noise in data.
 Isolation of patterns of interest from the background.
 Feature extraction:
 Finding a new representation in terms of features.
 Classification
 Using features and learned models to assign a pattern to
category.
 Post-processing
 Evaluation of confidence in decisions.
17

18. AN EXAMPLE OF PATTERN RECOGNITION
 Classification of fish into two classes: salmon and Sea Bass by
discriminative method
18
 Sorting incoming
fish on a conveyor
belt according to
species.
• Salmon
• Sea-bass

19. Steps of the process:
1. Capture image
2. Take measurements
 Length
 Width
 Number and shape of fins
 Position of the mouth, etc…
3. Make decision
19

20. 20
Pre-processing
Feature Extraction
Classification
“Sea Bass” “Salmon”

21.  Pre-Processing:
i. Image enhancement
ii. Isolate fishes from one another and from the background
iii. Finding the boundary of the fish
 Feature extractor:
i. Reduce the data by measuring certain features.
ii. To improve recognition, we might have to use more than
one feature at a time.
21
Fish x = [x1, x2]
Lightness width

22. DECISION/CLASSIFICATION BOUNDARIES ?
22

23. THANK YOU
23