American Journal of Intelligent Systems 2012, 2(1): 23-27

DOI: 10.5923/j.ajis.20120201.04

Pattern Recognition: an Overview

Vinita Dutt*,Vikas Chaudhry, Imran Khan

Bhagwant University, Ajmer Rajasthan, India

Abstract Pattern recognition has become more and more popular and important to us since 1960’s and it induces
attractive attention coming from a wider areas. In this paper Pattern recognition was introduced including concept, method,
application and integration. At the same time, ten definitions and more than ten methods of pattern recognition were
summarized. On the end, the structure and classification of PR and its related fields and application areas were introduced in
detail.
Keywords Pattern Recognition, Definition, Methods, Application

concerns the description or classification (recognition) of

1. Introduction measurements”[6]
1993(Srihari, Govindaraju) defined pattern recognition as
Pattern recognition is not unfamiliar with everyone, it has a discipline which learn some theories and methods to de-
a long history. PR is a subject researching object description sign machines that can recognize patterns in noisy data or
and classification method, it is also a collection of mathe- complex environment.[7]
matical, statistical, heuristic and inductive techniques of 1996(Ripley) outlined pattern recognition in his book:
fundamental role in executing the tasks like human being on “Given some examples of complex signals and the correct
computers. In a sense, PR is figuring out actual problems via decisions for them, make decisions automatically for a
mathematical methods. stream of future examples”[8]
2002(Robert P.W. Duin) described the nature of pattern
recognition is engineering; the final aim of Pattern recog-
2. The Definition of Pattern Recognition nition is to design machines to solve the gap between ap-
plication and theory.[9]
1973(Duda and Hart) defined the pattern recognition is a
2003(Sergios Theodoridis) Pattern recognition is a sci-
field concerned with machine recognition of meaning
entific discipline whose aim is the classification of the ob-
regularities in noisy of complex environments.[1]
jects into a lot of categories or classes. Pattern recognition
1977(Pavlidis) defined pattern recognition in his book:
is also a integral part in most machine intelligence system
“the word pattern is derived from the same root as the word
built for decision making.[10]
patron and, in his original use, means something which is
set up as a perfect example to be imitated. Thus pattern
recognition means the identification of the ideal which a
given object was made after.”[2] 3. The Research of Pattern Recognition
1978(Gonzalez, Thomas) defined pattern recognition as a Methods
classification of input data via extraction important features
from a lot of noisy data.[3] Pattern recognition undergoes an important developing
1985(Watanabe) said that pattern recognition can be for many years. Pattern recognition include a lot of
looked as categorization problem, as inductive process, as methods which impelling the development of numerous
structure analysis, as discrimination method and so on.[4] applications in different filed. The practicability of these
1990(Fukunaga) defined pattern recognition as” A methods is intelligent emulation.
problem of estimating density functions in a high- dimen-
sional space and dividing the space into the regions of 3.1. Statistical Pattern Recognition
categories of classes.”[5] Statistical decision and estimation theories have been
1992(Schalko ff) defined PR as“The science that commonly used in PR for a long time. It is a classical
method of PR which was found out during a long devel-
*Corresponding author:
[email protected] (Vinita Dutt)
oping process, it based on the feature vector distributing
Published online at http://journal.sapub.org/ajis which getting from probability and statistical model. The
Copyright © 2012 Scientific & Academic Publishing. All Rights Reserved statistical model is defined by a family of class-conditional
24 Vinita Dutt et al.: Pattern Recognition: an Overview

probability density functions Pr(x|ci) (Probability of feature one pattern. There are two main methods in structural pat-
vector x given class ci) In detail, in SPR, we put the fea- tern recognition, syntax analysis and structure matching.
tures in some optional order, and then we can regard the The basis of syntax analysis is the theory of formal lan-
set of features as a feature vector.[11]Also statistical pat- guage, the basis of structure matching is some special
tern recognition deals with features only without consider technique of mathematics based on sub-patterns. When
the relations between features. consider the relation among each part of the object, the
structural pattern recognition is best. Different from other
3.2. Data Clustering methods, structural pattern recognition handle with symbol
Its aim is to find out a few similar clusters in a mass of information, and this method can be used in applications
data which not need any information of the known clus- with higher level, such as image interpretation.
ters. It is an unsupervised method. In general, the method Structural pattern recognition always associates with sta-
of data clustering can be partitioned two classes, one is tistic classification or neural networks through which we
hierarchical clustering, and the other is partition clustering. can deal with more complex problem of pattern recognition,
such as recognition of multidimensional objects.
3.3. The application of Fuzzy Sets
3.6. Syntactic Pattern Recognition
The thinking process of human being is often fuzzy and
uncertain, and the languages of human are often fuzzy also. This method major emphasizes on the rules of compo-
And in reality, we can’t always give complete answers or sition. And the attractive aspect of syntactic methods is its
classification, so theory of fuzzy sets come into being. suitability for dealing with recursion. When finish cus-
Fuzzy sets can describe the extension and intension of a tomizing a series of rules which can describe the relation
concept effectively. among the parts of the object, syntactic pattern recognition
The application of fuzzy sets in pattern recognition which is a special kind of structural pattern recognition can
started in 1966, where the two basic operations– abstrac- be used.( in the middle of 1960’s,1978)[19]
tion and generalization were quite much aimed at by Bellan
et al.[12] Two principles proposed by Marr (1982) and 3.7. Approximate Reasoning Approach to Pattern
(Keller, 1995) which can be think as the general role of Recognition
fuzzy sets in PR.[13-16] The PR system based on fuzzy This method which uses two concepts: fuzzy applications
sets theory can imitate thinking process of human being and compositional rule of inference can cope with the
widely and deeply. problem for rule based pattern recognition. (Kumar S.Ray,
J.Ghoshal, 1996)[20]
3.4. Neural Networks
Neural networks is developing very fast since the first 3.8. A Logical Combinatorial Approach to Pattern
neural networks model MP was proposed since 1943, espe- Recognition
cially the Hopfield neural networks and famous BP arith- This method is presented, and works mainly in Spanish
metic came into being after. and Russian, which works with the descriptions of the
It is a data clustering method based on distance meas- objects. This approach can apply for both supervised pat-
urement; also this method is model-irrespective. The neural tern recognition and unsupervised pattern recognition.[21]
approach applies biological concepts to machines to recog-
nize patterns. The outcome of this effort is the invention of 3.9. Applications of Support Vector Machine (SVM) for
artificial neural networks which is set up by the elicitation Pattern Recognition
of the physiology knowledge of human brain. Neural net-
works is composed of a series of different,associate unit. In SVM is a relative new thing with simple structure; it has
addition, genetic algorithms applied in neural networks is been researched widely since it was proposed in the 1990’s.
a statistical optimized algorithms proposed by Holland SVM base on the statistical theory, and the method of
(1975)[17] SVM is an effective tool that can solve the problems of
NeurPR is a very attractive since it requires minimum a pattern recognition and function estimation, especially can
priori knowledge, and with enough layers and neurons, an solve classification and regression problem, has been ap-
ANN can create any complex decision region. plied to a wide range for pattern recognition such as face
detection, verification and recognition, object detection and
3.5. Structural Pattern Recognition recognition ,speech recognition etc.[22]
The concept of structural pattern recognition was put for 3.10. Using Higher-Order Local Autocorrelation
the fourth time (Pavilidis, 1977).[18] And structural pattern
Coefficients to Pattern Recognition
recognition is not based on a firm theory which relies on
segmentation and features extraction. Structural pattern In 2004, Vlad Popovici, present an efficient method us-
recognition emphases on the description of the structure, ing higher order autocorrelation functions for pattern rec-
namely explain how some simple sub-patterns compose ognition. The autocorrelation feature vectors reside in a
 American Journal of Intelligent Systems: 2012; 2(1): 23-27 25

high dimensional space, which one can avoid their com- These are mainly classification of PR system, whether
puting easily.[23] the system is successful mainly depends on his decision like
an expert or not.[25]
3.11. A Novel Method and System of Pattern
Recognition Using Data Encoded as Fourier series
and Fourier Space
It was put forward by (Randell. L Mills) in 2006. This
novel method anticipate the signal processing of an ensem-
ble of neurons as a unit and intends to simulate aspects of
brain which bring capabilities like pattern recognition and
reasoning that have not been produced with past ap-
proaches as neural networks.[24]

4. Pattern Recognition System

A pattern recognition system can be regarded as a proc-
ess that allows it to cope with real and noisy data. Whether
the decision made by the system is right or not mainly de-
pending on the decision make by the human expert.

4.1. The Structure of Pattern Recognition System

A pattern recognition system based on any PR method
mainly includes three mutual-associate and differentiated
processes. One is data building; the other two are pattern
analysis and pattern classification .Data building convert
original information into vector which can be dealt with by
computer. Pattern analysis’ task is to process the data Figure 1. the composition of a PR system
(vector), such as feature selection, feature extraction, data-
dimension compress and so on. The aim of pattern classi-
fication is to utilize the information acquired from pattern 5. Applications
analysis to discipline the computer in order to accomplish
the classification. It is true that application was one of the most important
A very common description of the pattern recognition elements for PR theory. Pattern Recognition has been de-
system that includes five steps to accomplish. The step of veloped for many years, and the technology of PR has been
classification/regression / description showed in fig1 is the applied in many fields such as artificial intelligence, com-
kernel of the system. puter engineering, nerve biology, medicine image analysis,
Classification is a PR problem of assigning an object to a archaeology, geologic reconnoitering, space navigation,
class, the output of the PR system is an integer label, such armament technology and so on. Detailed applications,
as classifying a product as “1” or “0” in a quality control such as below:
test. Pattern recognition applications
Regression is a generalization of a classification task, Pattern recognition is used in any area of science and en-
and the output of the PR system is a real-valued number, gineering that studies the structure of observations. It is now
such as predicting the share value of a firm based on past frequently used in many applications in manufacturing in-
performance and stock market indicators. dustry, healthcare, and the military. Examples include the
Description is the problem of representing an object in following.
terms of a series of primitives, and the PR system produces Optical character recognition (OCR) is becoming an in-
a structural or linguistic description. tegral part of document scanners, and is also used frequently
A general composition of a PR system is given below. in banking and postal applications. Printed characters can
now be accurately recognized, and the improving perform-
4.2. The Classification of Pattern Recognition System ance of automatic recognition of handwritten cursive char-
Rule based system acters has diminished significantly the need of human in-
Classical fuzzy teraction for OCR tasks.
System Bayesian system Automatic speech recognition is very important for user
Neural networks system interaction with machines. Commercial systems for auto-
Fuzzy neural networks systems matic response to flight queries, telephone directory assis-
26 Vinita Dutt et al.: Pattern Recognition: an Overview

tance, and telebanking are available. Often the systems are image analysis Automatic spectroscopy
tuned to a specific speaker for better recognition accuracy. • Bioinformatics
Computer vision deals with the recognition of objects as DNA sequences analysis
well as the identification and localization of their three- DNA micro array data analysis[29] Research of heredity
dimensional environments. This capability is required, for • Agriculture Output analysis Soil evaluating
example, by robots to operate in dynamic or unknown en- Extraction mineral characterization in coffee and sugar
vironments. This can be useful for applications ranging from [30]
manufacturing to household cleaning, and even for rescue • Geography
missions. Earthquake analysis Rocks classification
Personal identification systems that use biometrics are • Engineering
very important for security applications in airports, ATMs, Fault diagnosis for vehicle system
shops, hotels, and secure computer access. Recognition can Recognition of automobile Type
be based on face, fingerprint, iris, or voice, and can be Improve the safety performance of automobile
combined with the automatic verification of signatures and • Military affairs
PIN codes. Aviation photography analysis
Recognition of objects on earth from the sky (by satellites) Automatism Aim recognition
or from the air (by aeroplanes and cruise missiles) is called
remote sensing. It is important for cartography, agricultural
inspection, detection of minerals and pollution, and target 6. Related Fields
recognition.
Table 1. the related fields of PR
Many tests for medical diagnosis utilize pattern recogni-
tion systems, from counting blood cells and recognition of Machine learning Formal languages
Adaptive signal processing Cognitive sciences
cell tissues through microscopes to the detection of tumours
Artificial Neural networks Biological cybernetics
in magnetic resonance scans and the inspection of bones and Exploratory data analysis Structural Modeling
joints in X-ray images. Fuzzy and genetic systems Mathematical statistics
Many large databases are stored on the repositories ac- Detection and estimation theory Nonlinear optimization
cessible via the internet or otherwise in local computers. Robotics and vision Computational Neuroscience
They may have a clear structure such as bank accounts, a
weak structure such as consumer behaviour, or no obvious
structure such as a collection of images. Procedures for 7. Conclusions
finding desired items (database retrieval) as well as learning
In its broadest sense pattern recognition is the heart of all
or discovering structures in databases (data mining) are
scientific inquiry, including understanding ourselves and the
becoming more and more important. Web search engines
real-world around us. And the developing of pattern rec-
and recommender systems are two example applications
ognition is increasing very fast, the related fields and the
• Computer vision
application of pattern recognition became wider and wider.
The first vision system presented was supposing the
In this paper we expatiate pattern recognition in the round,
objects with geometric shapes and optimized edges ex-
include the definition of PR, the methods of PR, the com-
tracted from images.[26,27,28]
position of PR system, the related fields of PR and the ap-
• Computer aided diagnosis
plication of pattern recognition.
Medical imaging, EEG, EEG signal analysis
In addition, it is an important trend to use pattern
Designed to assist physicians, such as: X-ray mammog-
recognition on engineering; we should make efforts on this.
raphy
And pattern recognition scientists should pay attention to
Highlighting potential tambours on a mammogram
new technique of PR, and enlarge the application areas of
• Character recognition
PR.
Automated mail sorting, processing bank checks; Scan-
ner captures an image of the text;
Image is converted into constituent characters
• Speech recognition
Human computer interaction, Universal access; Micro- REFERENCES
phone records acoustic signal;
[1] Editorial,Advances in Pattern Recognition,Pattern Recogni-
Speech signal is classified into phonemes and words
tion Letters 26,395-398,2005
• Safety
Face recognition [2] T. Pavlidis. Structural Pattern Recognition. Springer Verlag,
Identifying fingerprints Berlin Heidelberg New York, 1977
• Astronomy [3] Gonzalez,R.C.Thomas,M.G .Syntatic Pattern Recognition:
Classifying galaxies by shape Astronomical telescope an Introduction ,Addison Wesley,Reading,MA,1978
 American Journal of Intelligent Systems: 2012; 2(1): 23-27 27

[4] Watanabe, Pattern Recognition: Human and Mechanical. casion of receiving the K.S. Fu prize, Pattern Recognition
Wiley, New York, 1985 Letters 24 ,1-7,2003
[5] K. Fukunaga. Introduction to statistical pattern recognition [20] KS Ray, J. Ghoshal, Approximate reasoning approach to
(2nd ed). Academic Press, Boston 1990 pattern recognition. Fuzzy Sets and Systems, 77:125-150,
1996
[6] RJ Schalkoff. Pattern Recognition: Statistical, Structural and
Neural Approaches. John Wiley & Sons,1992 [21] Jose Franciso martine-Thrinidad, The logical combinatorial
approach to pattern recognition, an overview through selected
[7] Srihari, S.N., Covindaraju, Pattern recognition, Chapman works, Pattern Recognition 34,741-751, 2001
&Hall, London, 1034-1041, 1993
[22] [22]Hyeran Byun, Seong-Whan Holland Lee, Applications of
[8] B. Ripley, Pattern Recognition and Neural Networks, Support Vector Machines for Pattern Recognition: A Survey,
Cambridge University Press, Cambridge, 1996 SVM 2002, LNCS 2388, pp. 213-236, 2002
[9] Robert P.W. Duin,Structural, Syntactic, and Statistical and- [23] Vlad Popovici*, Jean-Philippe Thiran, Pattern recognition
Pattern Recogition : Joint Iapr International Workshops Sspr using higher-order local autocorrelation coefficients, Pattern
2002 and Spr 2002, Windsor, Ontario, Canada, August Recognition Letters 25 ,1107-1113,2004
6-9,2002 Proceedings
[24] Novel method and system for pattern recognition and
[10] Sergios Theodoridis, Konstantinos Koutroumbas , pattern processing using data encoded as Fourier series and Fourier
recognition , Pattern Recognition ,Elsevier(USA)) ,1982 space, Randell L. Mills*, Engineering Applications of Ar-
tificial Intelligence 19 ,219-234, 2006
[11] Frank T.Allen*,Jason M.kinser, H. John Gaulfield, A neural
bridge from syntactic pattern recognition to statistical pattern [25] M.petrou, Learning in pattern recognition: some thoughts,
recognition ,Neural Networks 12 ,519-526 ,1999 Pattern Recognition Letters 22,3-13, 2001
[12] R.Bellman, R.Kalaba, L.A.Zadeh, Abstraction and pattern [26] Robert, L.G., 1965, Mahcine perception of three-dimensional
classification, J.Math. Anal. Appl. 13, Fuzzy sets and solids. Tippett,J.T. et al.(Eds), Optical and Electro Optical
systems (1996) Information Processing.MIT Press,Cambrige, MA, pp. 159-
197,1965
[13] Bezdek, J.C. and S.K. Pal, Eds, Fuzzy Models for pattern
recognition. IEEE Press New York, 1992 [27] Guzman, A. Decomposition of a visual scene into three di-
mensional bodies AFIPS Conference Proceedings. Thompl-
[14] Keller,J.M., Fuzzy sets in computer vision. Proc. VI IFSA son, Washington,DC, Vol 33, pp.291-304, 1968
World Congress, Sap Paulo, Brazil, Vol.I, 7-10, 1995
[28] M.Petrou, Learning in pattern recognition: Some thoughts,
[15] Marr, D.Vision. W.H. Freeman, San Francisco, CA, 1982 Pattern Recogniton Letters 22, 3-13, 2001
[16] Witold pedrycz, Special issue on fuzzy set technology in [29] Alan Wee_Chung Liew, Hong Yan,Mengsu Yang, Pattern
pattern recognition, Pattern Recognition Letters, 1996 Recognition techniques for the emerging field of bioinfor-
[17] Holland JH, Adaption in Natural and Artificial systems. matics: A review, Pattern Recognition, 2005
University of MICHIGAN press, Ann Arbor, 1975 [30] Pattern recognition applied to mineral characterization of
[18] Pavilidis, T., Structural Pattern Recognition, Springer-Verlag, Brazilian coffees and sugar-cane spirits, Mirian C. santos,
New York, 1977 Sherlan G.Lemos AnaRita ANogueira

[19] Theo Pavlids, 36 years on the pattern recognition front

Lecture given at ICPR’2000 in Barcelona, Spain on the oc-