Methods Of Pattern Recognition Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The human brain is built around pattern recognition and also human brains are excellent tools for analyzing patterns. The human brain and function of its different parts can be used in pattern recognition. Pattern Recognition is the human intelligence which stands deeply in different human activities. Human pattern recognition can be considered as a typical observation process which depends on knowledge and experience people already have. Pattern recognition means to a process of inputting pattern information and matching with the information in long tern memory and then recognizing the category which the pattern belongs to. Therefore the pattern recognition depends on people's knowledge and experience. The pattern recognition can be defined as follows,

Pattern recognition is a systematic discipline whose aim is the classification of the objects into a lot of categories. Pattern recognition is also a part in most machine intelligence system built for decision making.

The science that concerns classification of measurements.

Human have dominant capability to classify objects based on sensory input. They can easily read the document printed in different type fonts, including handwritten documents. For example ability is amazing because it often seems to require little Mindful effort.

Pattern recognition methods

Pattern recognition include a lot of methods which impelling the development of several application in different category. These methods are practically in intelligent emulation.

Statistical Pattern recognition

It is a classical method which was found out during a long developing process. It based on the feature which getting from probability and statistical model. Statistical pattern recognition only deals with features without consider the relations between features.

Data Clustering

The main of this method is to find out few similar clusters in mass of data not need any information of the known clusters. This method is a unsupervised method.

Neural Networks

The neural approach applies biological concepts to machine to recognize patterns. The main outcome of this method is the invention of artificial neural networks which is set up by the elicitation of the physiology knowledge of human brain.

Structural Pattern Recognition

The structural pattern recognition is not based on segmentation and features withdrawal. This method stress on the description of the structure, namely explain how some simple sup patterns create one pattern. Structural pattern recognition always deals with static classification or neural networks through which we can deal with more complex problem of pattern recognition.

Syntactic Pattern Recognition

This method focus on the rule of composition. This pattern recognition based on which is a special kind of structural pattern recognition.

Pattern Recognition systems

Pattern recognition system is the process that allows cope with real and noisy data. Whether the decision made by the system is right or not mainly depending on the decision make by human professional.

The structure of pattern recognition system

A pattern recognition system based on pattern recognition method mainly includes three different processes. The first 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. For example,

Feature Selection

Feature Extraction

Data dimension compression.

The main purpose of pattern classification is to operate the information acquired from pattern analysis to control the computer in order to bring about the classification. A very common description of the pattern recognition system that includes five steps to carry out. The step of classification, regression and description is most important part of the system.

Classification is pattern recognition problem of transferring an object to a class. Regression is a simplification of a classification task, and the output of the pattern recognition system is a real valued number such as predicting the share value of a firm based on past performance.

Description is the problem of representing an object in terms of a series of primitives and also the pattern recognition system produces a structural description. In the next page a general composition of a pattern recognition system is given.

Post Processing





Feature Extraction

Pattern acquiringObjects


Pattern Recognition Applications

Pattern recognition application has been developed for many years and the technology of pattern recognition has been applied in many fields such as

Artificial intelligence

Computer engineering

Nerve biology,

Medicine image analysis

Space navigation


Geological reconnoitering

Armament technology and so on.

A list of detailed pattern recognition applications are given below.

Computer Vision

The first vision system presented was assumption the objects with geometric shapes and optimized edges extracted from images.

Computer aided diagnosis

Medical imaging, EEG, X-ray mammography.

Character recognition

Automated mail storing, processing bank checks.

Speech Recognition

Human computer interaction, Universal access.


Face Recognition, Identifying fingerprints


Classifying galaxies by shape, Astronomical telescope image analysis, Automatic spectroscopy.


DNA sequences analysis, DNA micro array data analysis.


Earthquake analysis, Rocks classification.


Fault diagnosis for vehicle system, Recognition of automobile type.

Military Affairs

Aviation photography analysis, automation aim recognition.

Voice Recognition

Voice recognition or speech recognition is a type of behavioral biometric technology which is used to identify a user with the help of already stored voice recordings (templates) in the biometric device. Every individual has a random voice structure which based upon number of characteristics and frequencies. The voice recognition technology helps to provide security at homes as well as offices.

Algorithms of speech recognition can be characterized generally as pattern recognition approaches and acoustic phonetic approaches. The success of voice recognition has been obtained using pattern recognition methods. The use of these pattern recognition techniques were applied to the problems of isolated word recognition, connected word recognition and continues speech recognition.

Voice/Speech recognition systems can be divided into two main types. Pattern recognition systems compare patterns to trained patterns to conclude a match. Acoustic Phonetic systems use familiarity of the human body (for example speech production and hearing) to compare speech features such as vowel sounds. But today most modern systems focus on the pattern recognition approach because it combines nicely with current IT techniques to a higher accuracy.

Voice recognition is a multilevel pattern recognition task where the acoustical signals are examined and structured into a hierarchy of sub word units, words, phrases and sentences. Each level in this may provide additional temporal constraints. For example known word pronunciations or word sequences which can give back for bugs at lower level.

The Voice pattern recognition can be broken down into the following steps,

Audio recording detection.

Pre filtering.

Framing the data into a usable format.

Further filtering frame.

Comparison and matching (recognizing the sound)

Action (performing function associated with the recognized pattern)

Audio recording detection

This can be utilized in a number of ways. First it will compare the audio level with the simple recording. Sometimes the endpoint detection is harder because speakers tend to leave artifacts including breathing, sighing, teeth chatter and echoes.

Pre Filtering

This can be done in a variety of ways. Depending on other features of the recognition system. The most common method used in this is Bank of Filters method. This method utilizes a series of audio filters to prepare the sample and the analytical coding method which uses a forecast function to calculate errors. Spectral analysis is also used.

Framing the data into a usable format

This involves in separating the sample into a specific size. It also involves preparing the sample boundaries of analysis for example removing edge clicks.

Further filtering frame

This consists of time alignment and normalization.

Comparison and matching (recognizing the sound)

This method involves in comparing the current window with known samples. Frequency analysis, differential analysis, spectral distortion and time distortion methods. All these methods are used to generate accuracy match.

Finally the Actions can be just doing anything what the developer wants to.

The structure of a speech recognition system

First the system will get the raw voice/speech to utilize the signal analysis and then it will create voice frames to accomplish the voice acoustic analysis here the analysis will be done using acoustic models and also it will train segmentation. Accordion to the analysis the time alignment will be identified by using sequential constraints. Then with clear time alignment of the voice the word sequence will be recognized from the system.

Below the picture shows the sequence of the speech recognition system.

Raw Speech





Acoustic model

Frames score







Speech frames



Features of Voice Recognition

Voice recognition is facilitated by microphones, which record the voice samples and store them for later use or future use.

This technology is the only form of biometric, which makes use of physical qualities like shape of the vocal behavior as well as behavioral qualities like voice texture, pronunciation and others.

Automated recognition and verification processes are facilitated by this voice recognition technology in today's Information technology world.

The voice recognition devices designed to offer voice recorder, voice alerting system and emerging power supports.

Application of Voice Recognition

Voice recognition systems are installed to devices which doesn't have any other alternative biometric analysis than the voice. Today these devices (with voice recognition facilities) are applied in various places. Some examples are given below.

Modern cell phones are installed with voice recognition components to control the access of unfamiliar callers.

ATM cards are also making use of this technology to ensure that the authorized users use their cards to withdraw money from ATM terminals.

Door locks are designed with voice recognition technology to ensure that strangers are not allowed to pose threats to the security of house and family.

Computer access is restricted to authorized users with help of devices based on this technology.

Advantages of Voice Recognition

Voice recognition system offers involuntary door locking, in case the door is left unlocked by mistake. The mechanism and components are easily installed in devices like mobile phones, ATM cards and computers. Low cost technology and fast identification process are some other benefits of this technology.


The number of complexity problems of pattern recognition appears often to be overhelimg.Many of its problems can be solved if the pattern recognition system uses the following four steps in pattern recognition. They are,



Feature extraction


These pattern recognition steps and methods can be used in voice recognition systems and applications. As a result, voice recognition system getting more important now a days. It is used in a lot of area. For example mobile phones include voice recognition that allows utterances such as Cell Home. Another basic example is Windows speech recognition in windows vista users to communicate with their computers by voice. Mainly it was designed for people who want to limit their use of the computer input devices such as mouse, keyboard etc. So many pattern recognition algorithms and methods are used to develop these voice recognition applications. The main draw backs in voice recognition systems are the human speaker does not only communicate with speech or voice, But also with body signals, hand waving, eye movements etc. This information is completely missed by automatic speech recognitions. The problem is the Continues speech, this introduce a difficult problem in speech recognition. So there are many problems but doesn't this mean that it is too hard. There has been significant improvement in voice recognition applications and voice recognition application will continue to improve. One thing that should be investigated further is if the human speak differently to computers. So the voice recognition application should understand all the patterns in human voice culture and variations to become efficient applications and the human speaker can adapt to the recognition systems to increase the quality of voice recognition.