Escolar Documentos
Profissional Documentos
Cultura Documentos
HELD: Petition denied. RTC and CA affirmed. 4. The task of authentication module is to authenticate the
identity of the person who intends to access the system. The
person to be authenticated indicates his/her identity and
I. BIOMETRICS places his/her finger on the fingerprint scanner; a digital image
of his/her fingerprint is captured; minutiae pattern is extracted
A. GENERAL PROCESS of Identification from the captured fingerprint image and fed to a matching
The problem of resolving the identity of a person can be algorithm which matches it against the person’s minutiae
categorized into two fundamentally distinct types of problems templates stored in the system database to establish the
with different inherent complexities identity.
1. Verification (authentication) - refers to the problem of
confirming or denying a person’s claimed identity (Am I who I
claim I am?).
2. Recognition - (Who am I?) refers to the problem of
establishing a subject’s identity
System Architecture
b. ulnar/radial/double loops
Applications of biometrics
• Physical access
• Attendance & time control
2. Galton’s minutiae
• Products
a. dot & ridge endings/spur
• Professional services
b. short and long ridges
• Social services
c. island/lake
• Forensics
d. bifurcation/double bifurcation/trifurcation
• Immigration
e. opposed bifurcation/ridge crossing
• Personal documents
B. FINGERPRINTS
1. Classification
a. plain/tented arches
C. ODONTOMETRICS Maxillary means “of or attached to a jaw or jawbone, especially
Odontometrics is the measurement and study of tooth size. It the upper jaw.” The maxillae are the two maxilla bones
is used in biological anthropology and bioarchaeology to study forming the upper jaw and palate of the mouth. The two halves
human phenotypic variation. The rationale for use is similar to are fused at the intermaxillary suture to form the upper jaw.
that of the study of dentition, the structure and arrangement of (Wiki)
teeth. There are a number of features that can be observed in The mandible lower jaw, or jawbone (from Latin mandibula,
human teeth through the use of odontometrics. (Wiki) "jawbone") in the human, is the largest, strongest and lowest
bone in the face. It forms the lower jaw and holds the lower
teeth in place. (Wiki)
a. Incisors
Incisors (from Latin incidere, "to cut") are the front teeth
present in most heterodont mammals. They are located in the
premaxilla above and on the mandible below. Humans have a
total of eight (two on each side, top and bottom).
Adult humans normally have eight incisors, two of each type.
The types of incisor are:
• maxillary central incisor (upper jaw, closest to the
center of the lips)
• maxillary lateral incisor (upper jaw, beside the
maxillary central incisor)
• mandibular central incisor (lower jaw, closest to the
center of the lips)
The teeth are the hardest substances in the human body. • mandibular lateral incisor (lower jaw, beside the
Besides being essential for chewing, the teeth play an mandibular central incisor)
important role in speech. Parts of the teeth include:
• Enamel: The hardest, white outer part of the tooth. Children with a full set of deciduous teeth (primary teeth) also
Enamel is mostly made of calcium phosphate, a rock- have eight incisors, named the same way as in permanent
hard mineral. teeth. Young children may have from zero to eight incisors
• Dentin: A layer underlying the enamel. Dentin is depending on the stage of their tooth eruption and tooth
made of living cells, which secrete a hard mineral development. Typically, the mandibular central incisors erupt
substance. first, followed by the maxillary central incisors, the mandibular
• Pulp: The softer, living inner structure of teeth. lateral incisors and finally the maxillary laterals. The rest of the
Blood vessels and nerves run through the pulp of the primary dentition erupts after the incisors.
teeth.
• Cementum: A layer of connective tissue that binds Apart from the first molars, the incisors are also the first
the roots of the teeth firmly to the gums and jawbone. permanent teeth to erupt, following the same order as the
• Periodontal ligament: Tissue that helps hold the primary teeth, among themselves. (Wiki)
teeth tightly against the jaw.
A normal adult mouth has 32 teeth, which (except for wisdom
teeth) have erupted by about age 13:
• Incisors (8 total): The middlemost four teeth on the
upper and lower jaws.
• Canines (4 total): The pointed teeth just outside the
incisors.
• Premolars (8 total): Teeth between the canines and
molars.
• Molars (8 total): Flat teeth in the rear of the mouth,
best at grinding food.
• Wisdom teeth or third molars (4 total): These teeth
erupt at around age 18, but are often surgically
removed to prevent displacement of other teeth.
The crown of each tooth projects into the mouth. The root of
each tooth descends below the gum line, into the jaw.
(WebMD)
B. Flower
− The fibers radiating from the center are distorted (in one
or more places) to produce the effect of petals
− May have only one significant petal with the remainder of
the iris looking like a stream
D. IRIS PATTERNS − The fibers form rounded openings in the iris - these vary in
size, density and intensity
A. Stream
C. Jewel
In iris recognition, the identification process is carried out by
gathering one or more detailed images of the eye with a
sophisticated, high-resolution digital camera at visible or
infrared (IR) wavelengths, and then using a specialized
computer program called a matching engine to compare the
subject's iris pattern with images stored in a database. The
matching engine can compare millions of images per second
with a level of precision comparable to conventional
fingerprinting or digital fingerscanning.
E. FACIAL RECOGNITION
Facial Recognition
• History
o First developed in the 1960s, the first semi-
− Can be recognized by the presence of pigmentation or automated system for face recognition required
colored dots on top of the fibers the administrator to (1) locate features [eyes, ears,
− These dots (or jewels) can vary in color from light orange nose, and mouth] on the photographs before it
to black (2) calculated distances and ratios to a common
− They can also vary in size reference point, which were then (3) compared to
− The pigmentation is superimposed above the fibers reference data.
o 1970’s: Goldstein, Harmon, and Lesk used 21
D. Shaker specific subjective markers (hair color, lip
thickness) to automate the recognition.
§ Problem up to this point: measurements and
locations were MANUALLY computed.
o 1988: Kirby and Sirovich applied PRINCIPLE
COMPONENT ANALYSIS.
§ This is a standard linear algebra technique.
§ Showed that less than 100 values were
required to accurately code a suitably aligned
and normalized face image.
o 1991: Turk and Pentland – using the eigenfaces
techniques – discovered that the residual error
could be used to detect faces in images.
§ Enabled real-time automated face recognition
systems.
§ 2001: Applied in Super Bowl – surveillance
images were compared to a database of
− Identified by the presence of both flower-like petals in digital mugshots.
the fiber arrangement and pigmented dots or jewels o PRESENT: used to combat passport fraud,
− The distribution and density of flowers and jewels can vary support law enforcement, identify missing
substantially between the two eyes children, and minimize benefit/identity fraud.
• Predominant Approaches
o Geometric (feature based) – looks at
IRIS PATTERN RECOGNITION distinguishing features
o Photometric (view based) – statistical approach
Iris recognition is a method of identifying people based on that distills an image into values and compares
unique patterns within the ring-shaped region surrounding the
the values with templates to eliminate variances
pupil of the eye. The iris usually has a brown, blue, gray, or
• Popular Algorithms
greenish color, with complex patterns that are visible upon
o Principal Components Analysis (use of
close inspection. Because it makes use of a biological
eigenfaces)
characteristic, iris recognition is considered a form of biometric
verification.
§ Statistical approach for classifying
samples of unknown classes based on
training samples with known classes.
• Aims to maximize between-class
(i.e. across users) variance and
minimize within-class (i.e. within
user) variance.
§ See figure 2: where each block represents a
class – there are large variances BETWEEN
classes, but little variance WITHIN classes.
§ Limitation: when dealing with high
dimensional face data, LDA faces the small
sample size problem that arises where there
are a small number of available training
samples compared to the dimensionality of
§ Probe and gallery images must be the SAME the sample space.
SIZE and must first be NORMALIZED to line o Elastic Bunch Graph Matching
up the eyes and mouth of the subjects within
the images.
§ PCA approach is then used to reduce the
dimension of the data by means of data
compression basics and reveals the most
effective low dimensional structure of facial
patterns.
• This reduction in dimension REMOVES
INFORMATION THAT IS NOT USEFUL § Relies on the concept that real face images
and precisely decomposes the face have many non-linear characteristics that are
structures into orthogonal components NOT addressed by the linear analysis
known as EIGENFACES. methods – such as variations in illumination,
• Each face image may be represented as a pose, and expression.
weighted sum of the eigenfaces, which are § A Gabor wavelet transform creates a
stored in a 1D array. dynamic link architecture that projects
§ Probe image is compared against a the face onto an elastic grid.
gallery image by measuring the • The Gabor jet is anode on the elastic grid,
distance between their respective notated by circles on the image, which
feature vectors. describes the image behavior around a
§ Limitation: typically requires the full frontal given pixel.
face to be presented each time – else, image o It is a result of a convolution of the
results in poor performance. image with a Gabor filter – which is
§ Advantage: can reduce the data needed to used to DETECT SHAPES and to
identify the individual to 1/1000th of the date EXTRACT FEATURES using image
presented. processing.
o Linear Discriminant Analysis • Recognition is based on the
similarity of the Gabor filter
response at each Gabor note.
§ Limitation: the difficulty with LDA is the
requirement of accurate landmark
localizations – which can sometimes be
achieved by combining PCA and LDA
methods.
• Where applied?
o Facebook (Do you wish to tag <friend>)
II. BEHAVIOMETRICS 1. Types of forgery1
a. Tracing and printing - A traced forgery is created by
A. HANDWRITING tracing over a genuine signature. A forgery of this type can be
tied to the original, or master, signatures if that original
1. Graphonomics/handwriting movement analysis signature can be located.
b. Simulated - A simulated forgery is produced by the writer
Graphonomics is the interdisciplinary field directed towards who learns to mimic a genuine signature. It may or may not be
the scientific analysis of the handwriting process, product, and possible to identify the forger, depending on the extent to
other graphic skills. Researchers in handwriting recognition, which the suspect’s normal handwriting characteristics.
forensic handwriting examination, kinesiology, psychology, c. Blind - A freehand forgery represents the suspect’s normal
computer science, artificial intelligence, paleography and handwriting, with no attempt to mimic the style of the genuine
neuroscience cooperate in order to achieve a better signature.
understanding of the human skill of handwriting. Research in
graphonomics generally involves handwriting movement 2. Basis for determining authenticity (form, line quality,
analysis in one form or another. formatting & content)
When a sufficient amount of writing from two different people
Handwriting movement analysis is the study and analysis of the is closely examined, there are always identifiable differences.
movements involved in handwriting and drawing. It forms an Comparisons of writing samples take into consideration a wide
important part of graphonomics, which became established variety of handwriting characteristics including word and letter
after the "International Workshop on Handwriting Movement spacing, slant or slope, speed, pen position, use of
Analysis" in 1982 in Nijmegen, The Netherlands capitalization, embellishments, legibility, use of punctuation,
and proportion of letters and other attributes.
2. Holographs/autographs The fluency or smooth flow of the writing line can also indicate
the authenticity of the writing. The lack of smooth line
A holograph is a document written entirely in the handwriting quality with tremor, pen stops and hesitations may indicate
of the person whose signature it bears. forgery.
An autograph (from the Greek: αὐτός, autós, "self" and γράφω, 3. Indicators of forgery2 (slowness of execution, touching up
gráphō, "write") is a document transcribed entirely in the change of hold (pressure), wavering (tremor), uncertain
handwriting of its author, as opposed to a typeset document interrupted strokes, marked variation of angles)
or one written by an amanuensis or a copyist; the meaning General indications of non-genuineness may include the
overlaps with that of the word holograph. following:
a. Blunt starts and stops - The forger places the pen
3. Handwriting automations point in contact with the paper, and then starts
writing. When he is finished with the name or some
During the last few years the pattern analysis and machine
portion thereof, he stops the pen and lifts it from the
intelligence community has developed automation tools for
surface. This may cause an emphasized blunt start or
forensic document examination (FDE), in particular for
determining whether a given handwriting specimen can be ending where the pen was placed in contact with the
attributed to known writing. As with other expert systems, such surface. At times this contact is held so long that if
as for medical diagnosis, current automation tools are useful the pen contains a fluid ink it will wet the paper and
only as part of a larger manually-intensive procedure. Defining migrate outward from the contact point.
a computational approach for the overall problem not only
places these tools in context but also helps validate and There may be unnecessary and extraneous marks
improve existing manual procedures. We consider the caused by pen starts and stops. The writer may
standard work flow in FDE of handwritten items and annotate decide after putting his pen in contact with the paper,
the steps where automation is available or possible. A well- that it is in the wrong spot, picks it up and moves it to
known ransom note case is considered as an example, where a position considered more correct. Normally a
there are multiple questioned documents, testing for multiple signature’s starts and stops are much more dynamic.
writers of the same document, determining whether the writing The pen is moving horizontally before it contacts the
is disguised, known writing is formal while questioned writing is paper and is lifted at the end while still in flight. This
informal, etc. The findings for the particular ransom note case leaves a tapered appearance at the beginnings and
using the tools are given. Observations are made for
developing a more fully automated approach to FDE.
1
http://highered.mheducation.com/sites/0072564938/student_view0/chapter3/cha
pter_outline.html
B. FORGERY
2
http://www.questioneddocuments.com/services-offered-by-ndl/signatures-
forgery/
endings of names or letters. There are, of course, mere presence of these indicators does not mean that the
exceptions to this. signature under scrutiny is non-genuine, but should contribute
b. Pen lifts and hesitation - This is occasioned when to the overall determination as to genuineness. Alternately, the
the pen stops at an unusual point in the writing; signature devoid of these indications may not be assumed to
perhaps where a radical change in direction is about be genuine. The signature of an elderly individual may, for
instance, be expected to contain tremor and hesitation. If,
to take place or a new letter formation is about to be
however, the questioned signature appears to be written in a
started. This may take on the appearance of a small
fluid manner and/or on a higher skill level than what is
gap in the written line where one is not expected, or expected, the red warning flag should be waving. This
an overlapping of two ink lines where there should be occurrence may itself be indicative of non-genuineness. Often,
only one continuous line. a forger, because of an inherent high skill level in his writing,
c. Tremor - Because the creation of most forms of non- may produce a product that contains fewer indications of
genuine signatures are little more than drawings, the forgery than a counterpart with a lower skill-level.
Language comprehension
2. Speech Recognition is called verification or authentication. On the other hand,
Speech recognition (SR) is the inter-disciplinary sub-field of identification is the task of determining an unknown speaker's
computational linguistics which incorporates knowledge and identity. In a sense speaker verification is a 1:1 match where
research in the linguistics, computer science, and electrical one speaker's voice is matched to one template (also called a
engineering fields to develop methodologies and "voice print" or "voice model") whereas speaker identification
technologies that enables the recognition and translation of is a 1:N match where the voice is compared against N
spoken language into text by computers and computerized templates.
devices such as those categorized as Smart Technologies and
robotics. It is also known as "automatic speech recognition" From a security perspective, identification is different from
(ASR), "computer speech recognition", or just "speech to text" verification. For example, presenting your passport at border
(STT). control is a verification process: the agent compares your face
to the picture in the document. Conversely, a police officer
Some SR systems use "training" (also called "enrollment") comparing a sketch of an assailant against a database of
where an individual speaker reads text or isolated vocabulary previously documented criminals to find the closest match(es)
into the system. The system analyzes the person's specific is an identification process.
voice and uses it to fine-tune the recognition of that person's
speech, resulting in increased accuracy. Systems that do not Speaker verification is usually employed as a "gatekeeper" in
use training are called "speaker independent" systems. order to provide access to a secure system (e.g. telephone
Systems that use training are called "speaker dependent". banking). These systems operate with the users' knowledge
and typically require their cooperation. Speaker identification
Speech recognition applications include voice user interfaces systems can also be implemented covertly without the user's
such as voice dialing (e.g. "Call home"), call routing (e.g. "I knowledge to identify talkers in a discussion, alert automated
would like to make a collect call"), domotic appliance control, systems of speaker changes, check if a user is already enrolled
search (e.g. find a podcast where particular words were in a system, etc.
spoken), simple data entry (e.g., entering a credit card
number), preparation of structured documents (e.g. a In forensic applications, it is common to first perform a speaker
radiology report), speech-to-text processing (e.g., word identification process to create a list of "best matches" and
processors or emails), and aircraft (usually termed Direct Voice then perform a series of verification processes to determine a
Input). conclusive match
The term voice recognition or speaker identification refers to
identifying the speaker, rather than what they are saying. Speaker diarisation (or diarization) is the process of partitioning
Recognizing the speaker can simplify the task of translating an input audio stream into homogeneous segments according
speech in systems that have been trained on a specific to the speaker identity. It can enhance the readability of an
person's voice or it can be used to authenticate or verify the automatic speech transcription by structuring the audio stream
identity of a speaker as part of a security process. into speaker turns and, when used together with speaker
recognition systems, by providing the speaker’s true identity. It
3. Speaker recognition (speaker identification; speaker is used to answer the question "who spoke when?" Speaker
verification; diarization) diarisation is a combination of speaker segmentation and
speaker clustering. The first aims at finding speaker change
Speaker recognition is the identification of a person from points in an audio stream. The second aims at grouping
characteristics of voices (voice biometrics). It is also called together speech segments on the basis of speaker
voice recognition.There is a difference between speaker characteristics.
recognition (recognizing who is speaking) and speech
recognition (recognizing what is being said). These two terms With the increasing number of broadcasts, meeting recordings
are frequently confused, and "voice recognition" can be used and voice mail collected every year, speaker diarisation has
for both. In addition, there is a difference between the act of received much attention by the speech community, as is
authentication (commonly referred to as speaker verification or manifested by the specific evaluations devoted to it under the
speaker authentication) and identification. Finally, there is a auspices of the National Institute of Standards and Technology
difference between speaker recognition (recognizing who is for telephone speech, broadcast news and meetings
speaking) and speaker diarisation (recognizing when the same
speaker is speaking). Recognizing the speaker can simplify the
task of translating speech in systems that have been trained on
specific person's voices or it can be used to authenticate or
verify the identity of a speaker as part of a security process.
There are two major applications of speaker recognition
technologies and methodologies. If the speaker claims to be of
a certain identity and the voice is used to verify this claim, this