Face Recognition Technology

FACE

RECOGNITION

TECHNOLOGYSHRAVAN HALANKAR

GOA UNIVERSITY

ELECTRONICS DEPT.

O U T L I N E

1. Introduction

2. Biometrics

3. Why Face Recognition?

4. Implementation

5. How it works?

6. Strengths & Weaknesses

7. Major Challenges in FRT

8. Applications

9. Future of FRT………

I N T R O D U C T I O N

Everyday actions are increasingly being handledelectronically, instead of ink and paper or face to face.

This growth in electronic transactions results in greatdemand for fast and accurate user identification andauthentication.

Access codes for buildings, banks accounts and computersystems often use PIN's for identification and securityclearances.

Using the proper PIN gains access, but the user of the PINis not verified. When credit and ATM cards are lost orstolen, an unauthorized user can often come up with thecorrect personal codes.

BIOMETRICS technology may solve this problem .

BIOMETRICS?A biometric is a unique, measurable characteristic of a human

being that can be used to automatically recognize an individual

or verify an individual’s identity.

Biometrics system is an automated system of identifying a person based on person’s physical or behavioral characteristics.

TYPES OF BIOMETRICSFingerprints RetinaHand Geometry

Signature IrisFACE RECOGNITION

• Face recognition technology is the least intrusive and fastest biometric technology. It works with the most obvious individual identifier – the human face.

• It requires no physical interaction on behalf of the user.

• It can use your existing hardware infrastructure, existing cameras and image capture Devices will work with no problems

• It is only Biometric that allow you to perform passive identification in many environments. (e.g: Identifying terrorist in a busy Airport terminal.)

• It does not require an expert to interpret the comparison result.

ACCURATE!!

THE HUMAN FACE

• The face is an important part of who you are and how people identify you.

• In face recognition there are two types of comparisons :

This is where the system compares the given individual with who that individual says they are, and gives a yes or no decision.

This is where the system compares the given individual to all the Other individuals in the database and gives a ranked list of matches.

Verification Identification

ALL IDENTIFICATION OR AUTHENTICATION TECHNOLOGIES OPERATE USING THE FOLLOWING FOUR STAGES:

1)Capture

A physical or behavioral sample is captured by the system during Enrollment and also in identification or verification process

2)ExtractionUnique data is extracted from the sample and a template is created.

3)Comparison

The template is then compared with a new sample stored in the data base

4)Match /non match

The system decides if the features extracted from the new Samples are a match or a non match

IMPLEMENTATION

Input face image(Capture)

Face feature extraction

Feature Matching

Decision maker

Output result

Facedatabase

COMPONENTS OF FACE RECOGNITION SYSTEM

Enrollment Module

An automated mechanism that scans and captures a digital or an analog image of a living personal characteristics.

Another entity which handles compression, processing, storage and compression of the captured data with stored data .

Database

Also consists of a preprocessing system. In this module the newly obtained sample is preprocessed and compared with the sample stored in the database. The decision is taken depending on the match obtained from the database.

Verification Module

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

1. Distance between the eyes

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

1. Distance between the eyes

2. Width of the nose

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

1. Distance between the eyes2. Width of the nose

3. Depth of eye sockets

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

1. Distance between the eyes2. Width of the nose3. Depth of eye sockets

4. Structure of cheek bones

Every face has at least 80 distinguishable parts called nodal points. Some of them are:

1. Distance between the eyes2. Width of the nose3. Depth of eye sockets4. Structure of the cheek bones

5. Length of jaw line

A general face recognition software conducts a comparison of these parameters to the images in its database.Depending upon the matches found, it determines the result.This technique is known as feature based matching and it is the most basic method of facial recognition.

A 3D facial recognition model provides greater accuracythan the feature extraction model.

It can also be used in a dark surroundings and has a ability to recognize the subject at different view angles.

Using 3D software, the system Goes through a number of stepsto verify the identity of an individual.

Acquiring an image can be done through a digital scanning device.

Once it detects the face, the system determines heads position, size and pose.

The system then measures the curves of the face on a sub-millimeter scale and creates a template.

The system translates this template into a unique code.

The image thus acquired will be compared to the images in the data base and if 3D images are not available to the database, then algorithms used to get a straight face are applied to the 3D image to be matched.

Finally in verification, the image is matched to only one image in the database and the result is displayed as shown alongside.

The most commonly used unique feature for facial recognition is iris of the eye. No two human beings, even twins have exactly similar iris.

1. Eigen face or PCA (Principal Component Analysis)

Other method;

1. EBGM -Elastic Bunch Graph Method.-2D Image

2. 3D Face Recognition Method 3D Image

FACE FEATURE EXTRACTION METHODS

EIGEN FACE OR PCA (PRINCIPAL COMPONENT ANALYSIS)

• PCA, commonly referred to as the use of eigenfaces, this technique pioneered by Kirby and Sirivich in 1988.

• The PCA approach typically requires the full frontal face to be presented each time otherwise the image results in poor performance.

• The primary advantage of this technique is that it can

reduce the data needed to identify the individual to

1/1000th of the data presented

PCA-PRINCIPAL COMPONENT ANALYSIS(EIGEN FACE METHOD)

• 1.Create training set of faces and calculate the eigenfaces ( Creating the Data Base)

• 2. Project the new image onto the eigen faces.

• 3. Check closeness to one of the known faces.

• 4. Add unknown faces to the training set and re-calculate

• Face Image as I(x,y) be 2 dimensional N by N array of

(8 bit) intensity values.

• Image may also be considered as a vector of dimension N2.

( 256x256 image = Vector of Dimension 65,536 )

y Image T1=I1(1,1),I1(1,2)…I1(1,N),I1(2,1)…….., I1(N,N)

x

CREATING TRAINING SET OF IMAGES

• Training set of face images T1,T2,T3,……TM.-

• 1. Average Face of Image =Ψ = 1 ( ∑M Ti ) ; M –no. of images

M i=1

Ψ average face

• 2. Each Training face defer from average by vector Φ

Φi Eigen face

Each Image Average Image

Ti Ψ

Φi =Ti - Ψ

FACE IMAGES USING AS EIGEN FACES TRAINING IMAGES

-IMAGE MUST BE IN SAME SIZE-

Facedatabase

Elastic Bunch Graph Matching(EBGM)

• EBGM relies on the concept that real face images have many non-linear characteristics that are not addressed by the linear analysis methods .

• Such as variations in illumination(outdoor lighting vs. indoor fluorescents), pose (standing straight vs. leaning over) and expression (smile vs. frown).

• A Gabor wavelet transform creates a dynamic link architecture that projects the face onto an elastic grid.

• The Gabor jet is a node on the elastic grid, notated by circles on the image below, which describes the image behavior around a given pixel.

Elastic Bunch Graph Matching(EBGM)

• Recognition is based on the similarity of the Gabor filter response at each Gabor node.

• The difficulty with this method is the requirement of accurate landmark localization.

STRENGTHS AND WEEKNESSES OF FRT

STRENGTHS

It has the ability to leverageexisting image acquisitionequipment.

It can search against staticimages such as driver’slicense photographs.

It is the only biometric ableto operate without usercooperation.

WEEKNESSES

Changes in acquisition environmentreduce matching accuracy.

Changes in physiologicalcharacteristics reduce matchingaccuracy.

It has the potential for privacyabuse due to no cooperativeenrolment and identificationcapabilities.

PERFORMANCE OF FRT

False acceptance rates(FAR)

• The probability that a system will incorrectly identify an individual or will fail to reject an imposter.

• FAR= NFA/NIIA

• Where FAR= false acceptance rate

NFA= number of false acceptance

NIIA= number of imposter identification attempts

False rejection rates(FRR)

• The probability that a system will fail to identify an enrollee.

• FRR= NFR/NEIA

• Where FRR= false rejection rates

NFR= number of false rejection rates

NEIA= number of enrollee identification attempt

The only way to overcome this challenge is better equipment, i.e. basically , use of high tech cameras.It is very much essential for the system to catch the image accurately.

The only way to overcome this challenge is better ALGORITHMS for facial recognitions. If the systems are programmed for every possible permutation and combination of the image, an accurate match can be achieved.

APPLICATIONS

It is being estimated that facial recognition technology will be the backbone of all major security, home and networking service.

With the growth of social networking over the web, unbelievably accurate facial recognition algorithms and advanced equipment, a person’s face, no mater ageing or disguises or damage, can be recognized and data about that person can be produced.

Twins Recognition.

QUESTIONS ?

1.www.biometrics.gov2.www.wikipedia.com3.www.howstuffworks.com4.www.face-rec.org5.www.facedetection.com6.Recognizing Face Images with

Disguise Variations Richa Singh, Mayank Vatsa and Afzel NooreLane Department of Computer Science & Electrical Engineering, West Virginia

University,USA