Class 1, Group 2: Face & Finger Print Recognition


Uploaded by WafeekWahby on 23.04.2012

Transcript:
>> Chaithanya Kumar Karuna: So we are going to present
the Face and Finger Print Recognition.
>> Dr. Wafeek Wahby: Would you please put
the microphone always near to your mouth?
Like this.
Yes.
>> Chaithanya Kumar Karuna: So we are going to present
the face and finger print recognition
biometric technologies.
So--and this, the major topics are--so the first three topics
are going to be discussed by me.
And the next face recognition technology is going to be
discussed by Latha and then Chandra Mouli
and then conclusion by Sachin.
So coming to what is biometrics?
So as we know that "bio" means of life, it is a Greek word
and "metrics" means measurable.
So biometrics is an authentication technique
that relies on measurable physiological
and individual characteristics.
So physiological and individual characteristics are some
of the attributes.
So the attributes are some hands, face, or iris, retina,
are some of the attributes each an individual have.
So by using this, we can pump some
identification technologies.
So these are biometrics.
So why we go for biometrics?
So, it is because the authentication and authorization
are becoming necessary for many everyday actions like--so if you
want to pick a child from--your child from daycare or something
in airports, performing some financial transactions, or
accessing some facilities like entering into parks or if you
want to enter into a rec center you need to be authorized.
So in authorization or in authentication,
why we go for biometrics?
So this is because of these two main reasons.
The first one is biometrics provide high degree
of certainty.
So by using this, we have 100 percent--not exactly 100 percent
but most accurate results.
And the second thing is, these biometrics are directly
or indirectly related to cost saving.
So it reduces risk of financial losses.
So it is better to have 500 dollars biometrics rather than
losing some 1,000 dollars or 10,000 dollars valuable files.
So what are the types of biometrics?
There are two types, physiological and behavioral.
Physiologicals are measured at some points like finger, hand,
hand geometry or iris, retina.
This can be measured and they are going to be stay
with you entire, your lifetime.
And then coming to behavior, so this depends
on the state of mind.
So like signature, voice change, if you are signing 100 or 400
signs, your sign may change.
So coming to voice, your voice which may change depending on
your sadness or happiness.
So mostly, the physiological are more accurate than--
compared with behavioral.
So the working of or the working
of this almost biometric technologies are like this.
So there will be a database, enrolled database and then
acquisition, data acquisition and template, the score,
and the threshold and data.
And these are some of the major attributes.
So enrollment is the persons who are going to be identified.
They need to be enrolled previously.
So whenever he wants to enroll a previous identity,
you need to present the data, biometric data such
as fingerprint, his iris or his retina.
So the presented biometric data is converted into template.
So the system is going to be compared, the digital data,
not the image.
So it is converted into ones and zeros by using
these ones and zeros and enrolled,
already enrolled data.
The score is generated and the score is compared
with the threshold value.
This threshold value set by the administrator
of the biometric system.
So the scores are compared with the threshold value.
Your score is carried in threshold and is authenticated
and is uploaded to a user system or entering two systems.
If not, if less than, if score is less than threshold value,
then he will be rejected.
So this is the basic working thing of almost
all biometric system.
So now, the face recognition technology is going to be
discussed by Latha.
>> Latha Adalam: Hi friends.
This face biometric system, each and everyday we use face
for recognizing other humans and it uses features of
the face to verify or/and identify the individuals.
Even though we are in a crowd, face recognition will
identify the face.
In here, there were five steps for face, facial recognition.
First, it captures the image.
From that image, it will separate the face
and it extracts the features from that face.
And the features will be stored in a template.
And if he's the first--if he's enrolling for the first time,
it will store the template and he is enrolled.
And if he is coming for the second time, it will compare the
template which has taken already and the live features which is
taken now and it will declare the matches based on the
threshold similarity scores.
And it works, it uses high revolution cameras and
moderate lighting on the face.
While doing the face recognition, the user should
be close to the camera and proper angle will be settled
and darker and white skins can be more difficult to enroll.
Here, there were some features which does not change over time
like upper ridges of the eye sockets, areas under the--
around the cheekbones, sides of the mouth, nose shape,
position of major features relative to each other.
Difficulties in this face biometrics are, these are the
twins and this is father and son, they have similar features
which causes difficulty to identify.
And this is intra-class similarity here,
the same person.
If he changes the position or lighting,
or if he gives a big smile, then the features will be changed.
Challenges here are, if he gives a big smile, the
features will be different when compared to the normal face.
And if he has a mustache or facial hair, or coloring
chemicals, or if he had some makeup,
it will cause difficulties.
The template size is more for facial recognition
which is a challenge.
It is used in applications like logging into the system,
and mobile identification, and for check cashing.
That's it, my friend will discuss fingerprint.
>> Chandra Mouliyenamadala: Coming to fingerprint,
this is one of the oldest recognition technology compared
to the face recognition.
Fingerprint is nothing but a bunch of lens which has
a fixed pattern that are aligned with each other.
And these fingerprints lines are patterned and that they
never change in the lifetime of a person.
Fingerprints are usually classified based on the seven
criteria that are shown.
The crossover, core, bifurcation, ridge ending,
island, and delta and pore.
These are the finger ridge configurations
and the fingers cannot-- usually fingers can extract
these configurations from the input device.
These are the common fingerprint sensors that are usually used.
Our typical fingerprints sensors are optical, silicon-based
capacitive sensors, ultrasound and thermal.
Optical sensor usually use near infrared or UV light to
capture fingerprint image.
Silicon based capacitives are able to--
uses electrical pulses from the finger and ultrasound
uses ultrasound waves to capture the image.
And thermal sensor uses thermal--heating patterns
of the finger.
Fingerprints are mainly classified into five types.
These classifications are based on the fingerprint
lines alignment.
There are major three types of them.
There are whorl, right loop, left loop, tented arch,
and arch.
Fingerprint authentication system usually consists of
five stages, the image acquisition
from the sensory board, image processing
which removes some impurities in the image.
Feature extractions, extract the common features from
the fingerprints and template generation.
When the template is generated, it is compared with the stored
template that is already stored in the database
and an addition is made by a template matching algorithm.
Some of the advantage and disadvantage will be discussed
by my friend Sachin.
>> Sachin Kumar Mothkur: All right, coming to the
advantages of face.
It is easier to use the face recognition system.
It is not hardware based, we just need to install a software
and we need to have a basic camera to capture the image.
So the most important advantage is the subject cooperation
is not needed.
That is ability to operate without physical contact
or digital complicity.
So it is used in video surveillances for capturing
or identifying criminals.
Coming to the disadvantages of face.
So the disadvantage of face is mainly affected by
the environmental and also the lighting effects.
So, if the person is coming for a second time and is posing for
the camera, so it should be the same effect to you,
I mean, the physiological features should not change.
They may be changing due to the hairstyle, makeup, facial hair,
eyeglasses, hats or scarves.
Another disadvantage is the potential privacy abuse because
some of the users don't cooperate with administrator
on their time of enrollment and identification.
Advantages of finger.
Fingerprints are the most accepted and consistent
algorithm because they are very unique even in the--
even in twins.
And fingerprint has many standards developed in--
of all the biometrics because to achieve the interoperability
between the different vendor systems.
So, it is better to use a fingerprint with
password protection.
Coming to their disadvantages, it can be--the system can be
easily fooled, so easily copied or reproduced using the
publicly available documentation and easy to acquire materials.
So the same scanner used to authenticate one fingerprint
person and when another person comes, the previous mark
of the oldest fingerprint person is left there.
So, it can be easily faked.
So even if the fingerprint is not attached to the owner,
it will permit authentication.
Coming to the screen on the fingerprint scanners tend to
retain obstruction buildup of oil,
residue from user's fingerprints.
So, it is difficult to obtain a good quality image
if it is used continuously.
Some of the social consequences.
What happens when a biometric file is stolen?
Biometric file is similar to a normal file which can be hacked
so it contains identity of the individuals, so
it may lead to the misuse of the identity of the individuals.
Another is some of the biometric technologies are biased to
some minorities depending upon the race, gender, age
and color of the skin.
So it provides inconvenience when it is set up to access in
public areas such as parks.
So, one of the major challenges facing by the biometric systems
is to provide access to the disability persons who do not
have hands or retinas or eyes.
Conclusion, in the near future biometry will be used more for
needs of increased technology, so the government must
gain the trust of people.
So, in order to gain, they are to conduct some
awareness programs on the use of biometrics.
Face recognition is widely used biometry.
Face recognition uses the characteristics of
face to recognize individuals, it is easier to
identity the criminals.
Conclusion for fingerprint.
Fingerprints are used for identification essentially
because of their uniqueness and consistency.
Fingerprint recognition is considered as
a successful method of--among the all.
It is thought successful because everyone has a unique one
and easy to use.
You just need to place a finger on the scanner.
That's it, thank you.
>> Latha Adalam: My conclusion is every biometric
has its advantages and disadvantages.
Through this biometrics, we are just trying to increase
the security and it does not ensure the security.
I hope in future, they will definitely succeed in this field
by providing required and sophisticated resources.
>> Chaithanya Kumar Karuna: So coming to me, after 9/11
incidents, the biometric security is a major technology
to protect people from evil things.
So the major issue in biometric technology is [unclear dialogue]
many people tend to hack the systems and find loopholes
and making some evil things like 9/11 incidents.
If they--if you consider 9/11 incidents, the people who
participated in that, they used to have more than 11 to 12
secured licenses, so it's kind of a loophole in
the biometric technology, so it need to be rectified.
So for this, the people need to be matured in
using biometric technology.
Everyone must know what is the system they are using so,
all can--they can be protected.
So, everyone needs to be mature in using the biometric
technology so that there can be possibility for avoiding
loopholes in security.
>> Chandra Mouliyenamadala: There are a lot of
misconceptions about the biometric policy but right now
the people have started to bother about security
rather than their privacy.
And the government is also deploying large number of
biometric systems across the nation and worldwide
to improve the security parameters.
>> Dr. Wafeek Wahby: Any other conclusion?
There are none.
Okay, thank you.
[Applause]
Now any questions for this group?
Any questions?
I have one, yes, here is one.
>> Student #1: Have you heard about gait
recognition technology and is that becoming widely used now
or is that still one the developmental stages?
>> Chaithanya Kumar Karuna: Actually gait recognition
technologies are behavioral, you know, that can be varied.
So, the person's walking style may be changed.
If you consider a fat to lean lady, so there will be
a lot change in her walking style or guys.
Some guys may be lean and they became fat so that there will be
a huge walking style differences so,
but it is still implementing.
It may be used in, you know, malls or somewhere so that--
so the customer, the person sitting in front of this shop
may know of people coming into his shop, so it's still
the other thing but may not be more secured
than other iris or retina.
>> Student #1: Okay, thank you.
>> Dr. Wafeek Wahby: Other questions?
Other questions?
I have a question.
In your research, did you find any thread that ties for the one
person all the biometrics thing for the eye, iris,
their fingerprints and other stuff.
For the same person, is there any like a DNA thing?
>> Chaithanya Kumar Karuna: Yeah, we do have DNA biometric
technologies--technology, so--
>> Dr. Wafeek Wahby: So the person has the same thing
that appears in eye and fingerprint and--
>> Chaithanya Kumar Karuna: Yeah, it comes on that
multi-biometric technology, you know, using more--
>> Sachin Kumar Mothkur: But it is not easy to use that.
>> Latha Adalam: It is not yet implemented.
>> Dr. Wafeek Wahby: Okay.
Now, this database has to be collected and put together,
so not everybody is in it now.
>> Sachin Kumar Mothkur: Yeah.
>> Dr. Wafeek Wahby: All right?
Okay, are they having plans to collect from everybody
in the world?
>> Chaithanya Kumar Karuna: Yeah.
>> Sachin Kumar Mothkur: But for the face recognition,
with all their concern, they capture or, I mean the
profile pictures from Facebook or some social networking sites
[iunclear dialogue].
>> Dr. Wafeek Wahby: Okay, any other questions?
So give them a big round of applause.
[Applause]