1
Q
Biometric
A
It relates to a measurement or set of measurements of human beings that can be used to identify them in the context of authentication
2
Q
Biometrics can be divided into two types:
A
1)Physiological biometric
2)Behavioral biometric
3
Q
Physiological biometrics
A
- They do not normally change throughout a person’s life, at least not after that person has reached adulthood
-Examples: fingerprint, iris pattern
4
Q
Behavioral biometrics
A
- Do not normally change, but can change if the person put a sustained effort to change them
- Gait or way of signing one’s name
5
Q
Biometric security system main advantage
A
1) People automatically have their identity token with them
2) A biometric cannot be forgotten or left behind
6
Q
Biometric security system main drawback
A
Many biometrics are not secret and so can be imitated
7
Q
Gait
A
- It is the way someone walks
- It is a behavioral biometric
8
Q
Advantages of gait
A
1) Can be captured at a distance
2) Can be captured without the subject’s knowledge
9
Q
Disadvantages of gait
A
1) It is relatively difficult to capture, normally involving working on a video sequence
2) It is also computationally intensive and so not suitable for real-time authentication
10
Q
Fingerprint
A
- It has to do with the ridges and furrows on a fingertip
-It is a physiological biometric
11
Q
Advantages of fingerprint
A
1) The ridge pattern on a fingertip stays the same throughout a person’s life (though it can become less pronounced), unless the fingertip is damaged
2) It is also highly distinctive - even identical twins have somewhat different fingerprints
3) A fingerprint is easy to capture
12
Q
Disadvantages of fingerprint
A
1) An estimated 5% of people are not able to provide a fingerprint that is usable for security purposes due to an injury or worn-down ridges on the fingertip
2) It is also possible to lift a fingerprint off a surface and use it to produce a fake finger
13
Q
Iris pattern
A
- It is the pattern of the iris in the eye
- It is a physiological biometric
14
Q
Advantages of iris pattern
A
It is unique for each eye and for each person
15
Q
Disadvantages of iris pattern
A
1) The capture does require the person’s cooperation so that the iris is at the correct distance from the camera (iris images can be captured at a short distance)
2) For some subjects, the light that is necessarily shone on the eye can cause irritation
16
Q
Name signature
A
- The way someone signs their name, considering the way they hold the pen and the pressure exerted
- It is a behavioral biometric
(The signature itself is different from the way the hand moves the pen. The way the way the hand moves the pen has been studied recently with a view to capture it and using it in biometric security systems)
17
Q
Advantages of name signature
A
It could well prove as acceptable as signatures themselves
18
Q
Disadvantages of name signature
A
How people sign their names and the signature produced do tend to change with time
19
Q
Typing pattern
A
- The way someone logs into their computer
- This is the way the keys are pressed:
- The pressure in which each key is tapped
- The intervals between key taps
- A behavioral biometric
20
Q
Advantages of typing pattern
A
- It has the potential to replace signatures for computer-based activities
- Its capture does not require the subject’s cooperation, as the computer can be set up to capture it automatically at login
21
Q
Disadvantages of typing pattern
A
Underdevelopment
22
Q
Personal odor
A
- A physiological biometric
- Devices to capture and analyze odor are very expensive
23
Q
Face
A
- It is a set of measurements relating to various facial features
- It is a physiological biometric
24
Q
Disadvantages of face biometric systems
A
- Electronic equipment still has difficulty with facial recognition
- Facial characteristics tend to change with the effects of aging, facial expressions, disguises, different angles of the face to the camera and different lighting conditions
- Close relatives may have similar facial biometrics
- Currently, the use of this biometric requires the subject’s cooperation as they must stand a certain distance from the camera
25
Voice
- Can be considered as both behavioral and physiological biometrics
- Behavioral because it relates to how people speak
- Physiological, if it's the frequency components of the voice that are being measured, as these relate to the vocal tract
26
Advantages of voice biometrics
Voice recognition is useable at a significant distance
27
Disadvantages of voice biometrics
- Humans recognize voices much more readily than electronic equipment can
- a sore throat can render a voice hard for a human to recognize, hence, it can make it impossible for electronic equipment to recognize
- close relatives may have voices that are so similar that identification is difficult for electronic equipment.
- Background noise is a problem for these systems.
- A subject’s voice can be pre-recorded without their knowledge and used to fool the recognition system
28
Characteristics of biometrics
1) Distinctiveness
2) Permanence
3) Universality
4) Collectability
5) Acceptability
6) Performance
7) Circumvention
29
1) Distinctiveness
Indicates how well the biometric separates one individual from another
30
2) Permanence
Indicates how well the biometric remains the same over time
31
3) Universality
Indicates how commonly the biometric is found in humans
32
4) Collectability
Indicates how easy it is to measure the biometric and translate it into electronic format
33
5) Acceptability
Indicates the degree of acceptance of the biometric by users
34
6) Performance
Indicates the accuracy, speed, and robustness of the system using the biometric
35
7) Circumvention
Indicates how easy it is to fool the system that uses the biometric
36
Biometric security systems
are security systems that are able to recognize the identities of individuals based on one or more of their biometrics.
Such systems are often linked to or are part of an authorization system, which decides whether to allow or deny access to protected areas based on one or more tests to authenticate an identity
37
The architecture of biometric security systems consist of:
1) Sensors
2)feature extractor
3) Storage
4) Matcher
5) Decision maker
38
Block diagram of biometric security systems
see pg1
39
Sensors
1) Devices that capture the raw biometric data of users
2) They can require contact with the biometric input, or they could be contactless
3) They influence the system's convenience of use, acceptability, and reliability
4) They influence how secure the system is
40
Feature extractor
1) Takes the raw data from the sensors as input, extracts from it the key features and converts them into a digital representation called a template
2) It can be seen as a form of non-reversible compression
3) Some systems encrypt the templates so that the data is better protected
41
Aims of a feature extractor
1) To reduce the complexity of the original biometric sample
2) To decrease the resources required to store and process the biometric
3) To prevent the original biometric data from being reconstructed from the template
42
The storage
~ It is a place to hold templates
~ The templates enrolled in the storage need to be protected and backed up
~ If a remote server is used, a secure communication link between the security system and the server is needed
43
Examples of storage devices
1) Local storage medium such as a smart card
2) The storage facilities of a computer (a hard disk)
3) A remote server
The benefit of storing a database of templates is that multiple security systems can share the data
44
The matcher
~ Compares the template created from an input biometric sample (the template outputted from the feature extractor) with a stored template
~ It measures the similarity (or differences) between the two templates and produces a quantitative reference such as similarity value
45
The decision maker
~ Decides whether the input template and the stored template match, based on the similarity value from the matcher
~ The decision maker must be able to tolerate both Intra-class and Inter-class differences.
~ The sensitivity of the decision maker is controlled by one or more thresholds
~ The threshold is usually set by the system administrator based on the level of security required and operational statistics
46
Intra-class differences
The differences between the templates of the same individual as each template from the same individual is likely to be slightly different due to many non-repeatable variations
47
Inter-class differences
The differences between the templates of different individuals
48
Biometric security systems operation modes
1) Enrollment mode
2)Recognition mode
49
Enrollment mode
1) A user's biometric data is captured by the sensors and then passed to the feature extractor to generate a template
2) The template will subsequently be sent to the storage medium for safe keeping
~To combat the inter-class differences and enhance the system's reliability, the system usually enrolls multiple templates for the same individual
~Biometric templates are correctly associated with the user's identity
50
Recognition mode
1) A user's biometric data is captured by the sensors and then passed to the feature extractor to generate a template
2) The matcher compares the newly created biometric template with one or more stored templates
3) The decision maker then takes into account the similarity measures produced by the matcher
4) The decision maker decides whether the input template matches the stored templates for that individual
51
Identification
Refers to a process of identifying one individual from a population of individuals. It is therefore a one-to-many comparison process.
52
During an identification process:
1) A biometric sample is taken from the individual and converted to a template
2) This template is compared in turn with each template in a database
~If only one exact match is found, the subject is identified
~When no exact match is found, the system usually offers the best matches instead, along with similarity values
The identification process can take considerable time
53
Authentication
Refers to a process of confirming whether an individual is who he or she claims to be. It is therefore a one-to-one comparison process.
54
During an authentication process:
1) A 'live' biometric sample is taken from the individual and converted to a template
2) This template is compared with stored templates of the same individual
55
More on authentication:
~ Before the authentication, the user has to let the system know who they are, either by presenting something they have or something they know
~ An authentication process takes much less time to conduct and is suitable for real-time authentication applications
56
Explain how does a simple facial recognition system works
~ The shapes and geometries of human faces are all different.
~ By establishing sufficient reference points on the face and
measuring the lengths between the reference points, it is possible to create a template that uniquely represents an individual.
57
A facial recognition system is prone to errors, since sometimes a computer cannot accurately detect all reference points. What is an example of a solution for this issue?
Increase the number of measurements
~ If the number of measurements is increased, the differences between two templates from different individuals are likely to increase (Inter-class differences increase)
~ These extra measurements should help to distinguish one individual from another
~ However,
~ These extra measurements increase the time and effort involved in creating the template
~ If the number of measurements is increased, the differences between templates from the same individual are likely to increase as well (Intra-class differences increase)
58
In a fingerprint, the most important features are
The ridge configurations
59
Ridge configurations
are the way the ridge lines and the valleys between them are arranged
60
The configuration of the ridge lines can be analyzed at three different levels
1) Global level
2) Local level
3) Micro levels
61
At the global level, attention is paid to
singularities
62
Singularities
are regions where ridge lines form high-curvature shapes
63
Singularities are classified into three main types:
1) Loop (can be further broken down into left and right loops)
2) Delta (can be further broken down into normal arches and tented arches)
3) Whorl
Therefore, five major classes of singularities can be found from fingerprints
(Slide 34 for figure)
64
General notes on singularities
~ The patterns on the ten fingertips should all be different, but they might also have some similar features
~ Some fingertips have only one singularity, but some have two types of singularity on one fingertip
65
At the local level, attention is paid to
the ridge lines individually to identify the point where a ridge line is discontinuous
66
The discontinuity points are called
Minutiae (minutiae is the plural of minutia)
67
The most common types of minutiae are:
1) Termination
2) Bifurcation
3) Lake
4) Independent ridge
5) Island or point
6) Spur and crossover
(Figure on slide 35)
68
At the micro level, attention is paid to
the fine intra-ridge details
~ Inside each ridge there are tiny sweat pores whose size ranges from 60 to 250 μm (micrometers)
~ These pores line the ridge along its entire length
~ Pore information such as density, position, and shape is highly distinctive
However, this information is not widely used by current automatic recognition systems because the reliable detection and analysis of this microscopic information requires very high-definition scanners
69
Fingerprint sensors:
~ are specifically designed to capture details of the fingertip. The output of a fingerprint sensor is a raw image of the fingerprint.
~ are usually arranged in a two-dimensional array and protected by a transparent layer of glass or plastic.
70
The two main fingerprint sensor types are:
1) Optical sensors
2) Solid state sensors
71
How do optical sensors work
~ They capture images of fingerprints by using light-sensitive sensors to convert optical patterns into electrical signals
72
The principal types of sensing element used in optical sensors are:
~ Charge-coupled devices (CCDs)
~ CMOS-based image sensors (complementary metal oxide silicon)
73
Optical sensors work by:
1) Shining light on the fingertip, which is placed on the transparent sensing surface of the sensor
2) Detecting the light reflected back on to the light-sensitive sensors
~Ridges: are in contact with the sensing surface, so they either scatter or absorb the light, so they appear dark
~Valleys: are at a distance from the surface, so they allow the light to be reflected to the light-sensitive sensors, so they appear lighter
74
Pros and cons of optical sensors:
~ Optical sensors are generally able to capture high-quality fingerprint images
~ Optical sensors physical size is much bigger than solid-state sensors
~ Optical sensors cannot be used in devices such as notebook computers or flash drives
75
How do solid-state sensors work?
Solid state sensors consists in an all-in-one silicon chip with a two-dimensional sensory array placed directly on the chip.
When providing a fingerprint image, users touch the sensing surface of the chip directly.
Solid-state sensors convert thermal, capacitive piezoelectric or electric field information to electrical signals
76
What is the most common type of solid-state sensors used?
Capacitive sensors are the most common type employed, because of their simplicity and low cost
77
Compare between optical and solid-state sensor:
~ Solid-state sensors have a reduced physical size and cost compared to optical sensors
~ Both optical and solid-state fingerprint sensors can be designed to acquire images through users either touching them or sweeping their finger over them
78
Touch sensors:
~ Have a large enough sensing area to cover the whole fingertip
~ Can capture the fingerprint in one go
~ Are more expensive and difficult to fit into small portable devices because of their size
~ Are simpler to produce and easier to use
79
Sweep sensors:
~ Sweep sensors are usually only a fraction of the size of a touch sensor
~ A sweep sensor can only capture a 'slice' of the fingerprint at a time (users are required to sweep their finger over the sensing area)
~ Sweep sensors are cheaper and smaller in size
~ The sweep is required to be at a steady speed and pressure
~ Users require a degree of training and/or practice before they can master a smooth sweep action
~The sweep action 'clean up' the sensor on each sweep so that no latent fingerprint is left on the sensor (a problem suffered by touch sensors), which makes it more difficult for fraudsters to lift a latent fingerprint off a sensor in order to make a duplicate of it
80
T215B
flashcards
Decks in class (10)
# Cards
