Li S.Z., Jain A.K. (eds.) Encyclopedia of Biometrics

Подождите немного. Документ загружается.

Dental Biometrics. Figure 7 Extracted dental work contours with and without image enhancement. (a), (b) and (c)

Without enhancement. (d), (e) and (f) After enhancement.

Dental Biometrics. Figure 8 Dental Atlas of a complete set of adult teeth containing indices and classification labels of

the teeth.

220

Dental Biometrics

query and the database record. The distance between

the query and all the database record is used to ﬁnd the

closest match for a given query, and according to

the rank, an ordered list of candidate identities is

generated.

Performance Evaluation

To the authors’ knowledge, no public domain data-

bases are available to evaluate the performance of au-

tomatic dental matching. Hong and Jain [6] conducted

experiments on a dental X-ray database consisting of

29 PM subjects and 133 AM subjects [6]. There are

360 PM tooth sequences (810 teeth) and 1,064 AM

tooth sequences (3,271 teeth) in total. Figure 11 shows

the Cumulative Match Characteristics (CMC) curve

based on this database. The accuracy for top-1 retrieval

is 19/29 (66%), the accuracy for top-4 retrievals is

23/29 (79%), and the accuracy for top-13 retrievals

is 26/29 (90%). Most of the errors in retrieval

are attributed to change in the appearance of dentition

due to loss of teeth, matching different types of teeth,

and poor quality of AM and PM images.

Other Approaches

Other matching approaches have been attempted for

human identiﬁcation based on dental radiographs.

Hofer and Marana [13]usededitdistancetomatch

dental codes extracted from dental works in radiographs.

Nikaido et al. [14] proposed to use image r egistration

approach. Nomir and Abdel-M ottaleb [15] compared

Fourier descriptors of tooth contours and other features

based on gray level information of teeth in images.

Dental Biometrics. Figure 9 SVM/HMM model for the upper row of 16 teeth. The circles represent teeth, and the

number inside each circle is the tooth index. The squares represent missing teeth, and the number inside each square is

the number of missing teeth.

Dental Biometrics. Figure 10 Examples of successful registration of the dental atlas to (a) a bitewing image, (b)a

periapical image, and (c) an image with a missing tooth. In (c), teeth numbered as 12, 14, and 15 are correctly registered.

The missing tooth (number 13) is detected.

Dental Biometrics

221

Summary

Massive human disasters mak e it imperativ e to research

automatic dental identiﬁcation methods to identify

anonymous human remains. Dental radiographs contain

valuable clues that often is the only source of informa-

tion to identify victims. An overview of automatic iden-

tiﬁcation methods based on dental radiographs was

given in this entry. The accuracy and efﬁciency of current

approaches need to be further improv ed. A large data-

base of AM and PM radiographs needs to be c ollected

and made available to researchers to evaluate perfor-

mance of the automatic systems under development.

Related Entries

▶ Feature Extraction

▶ Forensic Applications, Overview

▶ Hidden Markov Models

▶ Support Vector Machine

References

1. O’Shaughnessy, P.: More than half of victims IDd. New York

Daily News (11 Sep. 2002)

2. Dental records beat DNA in tsunami IDs. New Scientist 2516, 12

(2005). Http://www.newscientist.com/article.ns?id=mg18725163.

900

3. Disaster v ictim identiﬁcation. Http://www.interpol.int/Public/

DisasterVictim/Guide

4. Fahmy, G., Nassar, D., Haj-Said, E., Chen, H., Nomir, O., Zhou, J.,

Howell, R., Ammar, H.H., Abdel-Mottaleb, M., Jain, A.K.:

Towards an automated dental identiﬁcation system (ADIS).

In: Proceedings of ICBA (International Conference on Bio-

metric Authentication), vol. LNCS 3072, pp. 789–796. Hong

Kong (2004)

5. Jain, A., Chen, H., Minut, S.: Dental biometrics: human identi-

ﬁcation using dental radiographs. In: 1Proceedings of Fourth

International Conference on AVBPA (Audio- and Video-based

Biometric Person Authentication), pp. 429–437. Guildford,

U.K. (2003)

6. Chen, H., Jain, A.K.: 1Handbook of Biometrics, chap. on

Automatic Forensic Dental Identiﬁcation. Springer, Berlin (2007)

7. Adams, B.: The diversity of adult dental patterns in the United

States and the implications for personal identiﬁcation. J. Forens.

Sci. 48(3), 497–503 (2003)

8. Chen, H.: Automatic forensic identiﬁcation based on dental

radiographs. Ph.D. thesis, Michigan State University (2007)

9. Chen, H., Jain, A.: Dental biometrics: Alignment and matching

of dental radiographs. IEEE Trans. Pattern Analy. Mach. Intell

27(8), 1319–1326 (2005)

10. Nomir, O., Abdel-Mottaleb, M.: A system for human identiﬁca-

tion from X-ray dental radiographs. Pattern Recogn. 38(8),

1295–1305 (2005)

11. Mahoor, M.H., Abdel-Mottaleb, M.: Classiﬁcation and number-

ing of teeth in dental bitewing images. Pattern Recogn 38(4),

577–586 (2005)

12. Jain, A., Chen, H.: Registration of dental atlas to radiographs

for human identiﬁcation. In: Proceedings of SPIE Conference on

Biometric Technology for Human Identiﬁcation II, vol. 5779,

pp. 292–298. Orlando, Florida (2005)

Dental Biometrics. Figure 11 Cumulative matching characteristics (CMC) curves for subject retrieval in a database of

133 subjects [6].

222

Dental Biometrics

13. Hofer, M., Marana, A.N.: Dental biometrics: Human identiﬁca-

tion based on dental work information. In: Proceedings of

Brazilian Symposium on Computer Graphics and Image Proces-

sing (SIBGRAPI), pp. 281–286 (2007)

14. Nikaido, A., Ito, K., Aoki, T., Kosuge, E., Kawamata, R.: A dental

radiograph registration algorithm using phase-based image

matching for human identiﬁcation. In: Proceedings of Interna-

tional Symposium on Intelligent Signal Processing and Commu-

nications, pp. 375–378 (2006)

15. Nomir, O., Abdel-Mottaleb, M.: Human identiﬁcation from

dental X-ray images based on the shape and appearance of the

teeth. IEEE Trans. Inform. Forens. Sec 2(2), 188–197 (2007)

Dental Identification

▶ Dental Biometrics

Deployment

The engineering of technological ideas and devices to

become useful by-product in real-industry environ-

ments for operators and users.

▶ Biometric Applications, Overview

Depth of Field (DOF)

Depth of Field (DO F) is the distance in front of and

beyond the subject that appears to be in focus. It

depends on the camera lens.

▶ Face Device

▶ Iris on the Move™

Dermis

Dermis is the inner layer of the skin. It varies in

thickness according to the location of skin. It is

0.3 nm on eyelid while 3.0 on back. It is composed of

three types of tissues which are present throughout the

layers named as: (1) collagen, (2) elastic tissue, and (3)

reticular ﬁbers. The two layers of the dermis are the

papillary and reticular layers. The upper, papillary

layer, which is the outer layer, contains a thin arrange-

ment of collagen ﬁbers. The lower, reticular layer is

thicker and made of thick collagen ﬁbers that are

arranged parallel to the surface of the skin.

▶ Anatomy of Hand

DET Curves

Detection Error Tradeoff curves are ROC (receiver

operating characteristic) type curves showing the

range of operating points of systems performi ng detec-

tion tasks as a threshold is varied to alter the miss

and false alarm rates and plotted using a normal devi-

ate scale for each axis. DET curves have the property

that if the underlying score distributions for the

two types of trials are normal, the curve becomes a

straight line. They have been widely used to present the

performance characteristics of speaker recognition

systems.

▶ Speaker Databases and Evaluation

Detector – Extractor

Traditionally, the term detector has been used to refer

to the tool that extracts the features from the image,

e.g., a corner, blob, or edge detector. However, this only

makes sense if it is a priori clears what the corners,

blobs, or edges in the image are, so one can speak of

‘‘false detections’’ or ‘‘missed detections.’’ This only

holds in the usage scenario where features are seman-

tically meaningful, otherwise extractor would probably

be more appropriate. The term detector is however

more widely used.

▶ Local Image Features

Detector – Extractor

223

Diffraction Limit

In optics, a fundamental limit on the resolution that

can be obtained with an imaging system. The limit

depends on the wavelength of the light used, the size

of the aperture of the optical system and the distance

between the optical system and object being imaged.

▶ Iris on the Move

Diffuse Reflection

Diffuse reﬂection is the reﬂection of light from a rough

surface. When a bunch of parallel light reach the rough

surface, the reﬂected light will be spread in all direc-

tions. It is the complement to specular reﬂection.

▶ Skin Spectroscopy

Digital Watermarking

▶ Iris Digital Watermarking

Digitizer

▶ Digitizing Tablet

Digitizing Tablet

SONIA GARCIA-SALICETTI,NESMA HOUMANI

TELECOM SudParis, Evry, France

Synonyms

Digitizer; Graphic tablet; Touch tablet; Tablet

Definition

A di gitizing table t is a sensitive input device that con-

verts a hand-drawn trajectory into a digital on-line

form, which is a sequence. This hand-drawn trajectory

may be a signature input, handwriting, or hand-drawn

graphics. A digitizing tablet usually consists of an elec-

tronic tablet and a pen or a stylus. When the electronic

tablet communicates with the pen, it is said to be

‘‘Active’’ and in this case the pen contains an electronic

circuit; otherwise the tablet is said to be ‘‘Passive.’’ In

some cases, the digitizer only consists of an electronic

(Active) pen, used on either standard or special paper.

Active

▶ digitizing tablets sample the pen trajectory at

regular time intervals (around 10 ms), generating a

time stamp and associated time functions; passive

digitizing tablets require dedicated acquisition soft-

ware to retrieve the sequence of time stamps and asso-

ciated time functions. When digitizing human pen

input, the resulting output may have different forms

according to the type of digitizer used. Active digitizing

tablets capture a sequence of time functions, including

pen position , pen pressure, and pen inclination, while

passive digitizing tablets, with acquisition software,

only allow a time stamp and the position of the stylus

on the tablet to be captured. In the special case of

electronic pens, the digitizer may capture some of the

previously mentioned time functions plus some

others, such as pen acceleration.

Introduction

Digitizing tablets available nowadays can be based on

electromagnetic technology (Active digitizing tablets),

touch screen technology (Passive digitizing tablets),

hybrid technology combining both (respectively called

‘‘Active mode’’ and ‘‘Passive mode’’), or, ﬁnally, in the

case of digitizers consisting only of an electronic pen

(no tablet, just an Active Pen and a sheet of standard

or special paper) on a variety of principles (mechani-

cal, optical). Each kind of digitizing tablet has been

discussed here.

Active Digitizing Tablet:

Electromagnetic Technology

This is the technology of choice in biometric applica-

tions related to online signature veriﬁcation and writer

224

Diffraction Limit

authentication by online handwriting, because it is the

technolog y with the highest resolution and accuracy at

the acquisition step.

In this case, the digitizing tablet is based on

▶ elec-

tromagnetic resonance technology [1] and it contains

electronics external to the touched surface. An active

digitizing tablet consists of a sensitive acquisition sur-

face incorporating a large number of overlapping loop

coils arranged horizontally and vertically, and a special

pen containing an inductor-capacitor circuit (LC cir-

cuit). Electromagnetic resonance t hen allows informa-

tion exchange between the tablet and the pen in the

following way: the tablet generates a magnetic ﬁeld,

which is received by the LC circuit in the pen, and then

the pen’s resonant circuit makes use of this energy to

return an electromagnetic signal to the tablet [1]. Here,

the horizontal and vertical wires of the tablet operate as

both transmitters and receivers of magnetic signals.

This exchange of information allows the tablet to de-

tect the position, pressure, and inclination of the pen.

Using electromagnetic resonance technology in

particular allows the pen position to be sensed without

the pen even having to touch the tablet (ability to

hover) and also the user’s hand may rest on the ﬂat

acquisition surface without affecting the acquisition

process (the capture of the time functions of position,

pressure, and inclination of the pen).

Active digitizing tablets are of two types: one

in which the tablet powers the pen, thus avoiding the

use of batteries in the pen (as in the case of well-known

digitizing tablets on the market, for example, Wacom

digitizers [1]) and the other in which the pen requires

batteries (as in the case of other vendors, such as

AceCad or Aiptek).

The

▶ sampling frequency of Active digitizing

tablets may be tuned for acquisition. For example, in

Wacom Intuos2 A6 USB tablet, the sampling frequency

of the hand-drawn signal can reach 200 Hz and is

frequently set around 100 Hz.

Passive Digitizing Tablets:

Touch-Screen Technology

In this case, all the electronics are inside the digitizing

tablet, based on touch-screen technology, and are acti-

vated by a non-sensitive stylus. Passive digitizing tablets

are integrated into other multifunction devices like PCs

(such as a Tablet PC), or handheld devices, such as

Personal Digital Assistants (PDAs) or Smartphones.

Speciﬁc acquisition software is required in these digiti-

zers in order to retrieve the sequence of time stamps

and associated time functions corresponding to the

hand-drawn trajectory on the Touch Screen. Passive

digitizing tablets allow fewer time functions (only a

time stamp and the associated pen position on the

Touch Screen), than Active digitizing tablets to be cap-

tured. Also, spatial resolution is variable in these digitiz-

ing tablets and is less precise than that obtained in

Active digitizing tablets.

There exist several types of touch-screen technology

(resistive, capacitive, infrared, Surface Acoustic Wave,

and Near Field Imaging) but the two mostly used for

online signature capture and more generally for online

handwriting recognition, that is, resistive and Pen-

Touch Capacitive technology, are discussed here.

Resistive Passive Digitizing Tablets

The most widely used Passive digitizing tablets today

are based on resistive technology [2]. This technology

can be summarized by the fact that a touch on the

screen generates a tension (a voltage) at a localized

point that is at the position of touch [2]. In a Resistive

Touch Sensor, there are two upper thin metallic con-

ductive layers separated by a thin space (often ﬁlled

with tiny spacer dots) in between two resistive layers

(Fig. 1). The Resistive Sensor is mounted above a

▶ liquid crystal display (LCD) to sense pressure. Pres-

sure from using either a stylus or a ﬁnger bends the

upper conductive layer, which is ﬂexible, producing an

electrical contact between the two conductive layers

received by the LCD. A controller locates the pen by

measuring the tension between the point of contact

and the edges or corners of the touch screen. Resistive

Sensors cannot distinguish pens from ﬁngers and do

not have the hover ability of Active digitizing tablets

(cannot detect proximity of the pen or ﬁnger without

actual pressure).

Pen-Touch Capacitive Passive Digitizing

Tablets

This technology can be summarized by the fact that

a touch with a tethered stylus (special stylus with a

conductive tip) reduces electrical charges on the

Digitizing Tablet

225

screen [2]. A layer that stores electrical charges is

placed on the glass panel of the monitor. When the

stylus touches the screen, some of these charges are

transferred to the stylus and thus, the charge on the

capacitive layer decreases. This decrease is measured in

circuits located at each corner of the glass panel, and

the information relayed by the computer to the touch

screen driver software to determine the coordinates of

the touch.

This technology has three modes: the screen can be

set to respond to ﬁnger input only, pen input only, or

both. The pen stylus is used in particular for signature

capture and offers online handwriting recognition

facilities. One popular touch screen using this technol-

ogy is ClearTek II produced by 3M Touch Systems [3],

sometimes integrated in monitors (LCDSA121-PEN-

S-OF [4]); other examples of this technology are

Apple PDAs, such as iPhone and iPod (in particular,

Songtak Technology CoLtd [5] designed a special

stylus with a conductive tip for iPhone and iPod).

Hybrid Digitizing Tablets: Active Mod e

and Passive Mode

Some digitizing tablets are hybrid as they can operate

in both Active and Passive modes. There are cases in

which such digitizing tablets combine capacitive and

electromagnetic technologies, such as the ClearPad and

Spiral sensors of Synaptics [6] (one of the leaders in

capacitive touch-sensing technology) and others in

which resistive and electromagnetic technologies are

combined, such as the Tablet-PC Sahara slate PC

i440D [7]. These hybrid digitizing tablets can distin-

guish pen input from ﬁnger input since, when there is a

magnetic ﬁeld, what produces the touch is an Active

Pen. Microsoft’s Tablet PCs belong to this category of

hybrid digitizing tablets.

Digitizing Tablets with Active Pen Only

In this case, the digitizing tablet is completely mobile,

since it consists of a sensitive pen, which is like a normal

ballpoint pen in shape as well as grip. Such devices

even use ink for writing but, by means of different

principles, a sequence of data from the hand-drawn

signal is stored in the memory of the device.

There are different categories of Active Pens,

according to the underlying principles allowing the

acquisition of a sequence of data from a hand-drawn

signal. The ﬁrst category is based on a

▶ piezoelectric

element that transforms a mechanical force into an

electrical signal (voltage); this category is represented

for instance by the ‘‘Marking Device’’ [8]. Another

category is based on

▶ strain gauges that transform

their deformation (strain) into a change in an electrical

signal, usually a resistive circuit; this category is repre-

sented by the SmartPen [9]. The third category is based

on optical sensors [10–15].

In the Electronic pen [8], the Marking Device

includes a pressure sensor and two acceleration sen-

sors. The pressure sensor is coupled to the tip of the

pen. The acceleration sensors adjacent to the pen tip

sense acceleration of the tip in two directions. These

sensors are based on piezoelectric transducers; for

example, the pressure sensor has at least one pair of

electrodes coupled to a piezoelectric element that is

compressed as a result of the pressure exerted on the

tip. The piezoelectric tr ansducer then conveys the

resulting compression into an electrical signal. From

the sequence of pressure values, pen acceleration and

Digitizing Tablet. Figure 1 Principle of resistive touch-screen technology.

226

Digitizing Tablet

temporal information related to the sampling of the

acceleration sensors, the Marking Device extracts other

features, such as speed, position, and angular informa-

tion about the acceleration vector [8].

In the second category, a mechanical force is trans-

formed into an electrical signal, by the use of strain

gauges. An example of this technology is the SmartPen

[9], made by the LCI Computer Group. The pen con-

tains micro-electromechanical sensors, one sensing

both force (pressure) and acceleration, the other sens-

ing tilt, and a radio transmitter to send the informa-

tion to a computer. The LCI-SmartPen serves to

authenticate the signer of a document through the

dynamics of the signature. While the signer writes,

micro-sensors measure the inclination and pressure of

the pen and the acceleration of the pen tip in

three directions by means of a ring-shaped deformable

aluminium structure provided with strain gauges.

The information is processed and then transmitted

to a computer. The position of the pen and the pen tip

velocity are calculated through measurements obtained

from the force/acceleration sensor and the tilt sensor.

In the third category, optical principles are used to

acquire a data sequence from the hand-drawn signal.

The ﬁrst type of Active Pen in this category relies

on a digital camera integrated in the pen and special

paper; this is the Anoto Active Pen (of Anoto AB

Company) [10, 11]. When writing with the Active

Pen on Anoto paper, which contains a special pattern

of numerous black dots, digital snapshots are taken

automatically by an integrated digital camera (more

than 50 pictures per second), a nd the dots of the

written pattern are illuminated by infrared light,

making them visible to the digital camera. In this

way, a sequence of information (timing, coordinates,

etc.) is captured from the hand-drawn signal. With the

very same principle, it is found that the Sony Ericsson

Chatpen [11 ] and the Logitech io pen [12] also require

the Anoto special paper.

The second ty pe of Active Pen in this category relies

▶ diodes; an example is the V-Pen using an Optical

Translation Measurement sensor (OTM sensor), in-

cluding a laser diode, detectors, and optics integrated

into a small transistor-style package [13]. The laser

diode shines laser beams on to the writing surface

and the OTM sensor analyzes how movements of the

pen affect the reﬂected wavelengths [13]. The relative

motion in three dimensions is measured based on the

Doppler Effect – the change in frequency and wave-

length of a wave as perceived by an observer moving

relative to the source of the waves. The V-Pen performs

signature recognition by combining the dynamic char-

acteristics of the signing action to the signature shape.

Other Active Pens in this category have appeared as

research prototypes [ 14, 15]. The ﬁrst measures pen

pressure (force) in three dimensions by Optical prin-

ciples [14]. The pressure sensor uses a ﬂexible element

in combination with an optical displacement sensor

based on

▶ light emitting diodes (LED) and photo-

diodes. The pressure on the pen tip is applied to a

ﬂexible structure that deforms. A plate, placed between

the infrared LED and the photodiode, moves together

with the deformed structure (Fig. 2). Its movement

changes the amount of incoming light on the photodi-

ode and, therefore, the electrical current of the Photo-

diode (that corresponds to a pressure value). In a

new device [15], pen inclination is measured by

means of a laser diode used as a light source. The

laser beam is reﬂected by a mirror mounted on the

end of the ink rod. The reﬂected beam is guided to a

Digitizing Tablet. Figure 2 Principle of the optical force

sensor.

Digitizing Tablet. Figure 3 Principle of the optical force

sensor.

Digitizing Tablet

227

Quadrant Photodiode (QPD) via a cube beam splitter

(Fig. 3). When writing, the mirror attached to the ink

rod is tilted in X and Y directions and the path of the

reﬂected beam changes. Consequently, the optical spot

on the photodiode moves, causing variations in the

output currents of the QPD cells. The sampling fre-

quency of the output voltages is set at 200 Hz and the

resulting sequence of voltages is interpreted as pen

inclination information.

Related Entries

▶ Acceleration

▶ Altitude

▶ Azimuth

▶ Biometric Sensor and Device, Overview

▶ Feature Extraction

▶ On-Line Signature Veriﬁcation

▶ Pen Inclination

▶ Pen Pressure

▶ Position

▶ Sensors

▶ Signature Features

▶ Signature Recognition

▶ Timing

▶ Velocity

References

1. http://www.wacom.com/productinfo/. Accessed 6 Dec 2007

2. Synaptics Incorporated, New Touch Screens Improve Hand-

held Human Interface. Synaptics White Paper WP-12, P/N

507–000003, Rev A. Synaptics Incorporated, San Jose California

3. solutions.3m.com/wps/portal/3M/en_US/3MTouchSystems/TS/.

Accessed 14 Dec 2007

4. www.touchscreens.com/lcdsa121-pen-s-kr.html. Accessed 14 Dec

2007

5. www.songtak.com.tw/. Accessed 6 Dec 2007

6. www.synaptics.com/onyx/. Accessed 6 Dec 2007

7. www.tabletkiosk.com/products/sahara/i400s_pp.asp. Accessed 20

Dec 2007

8. O’Connor, M., Vannier, D.S.: Electronic pen device. US Patent

6,188,392, 13 Feb 2001

9. Reynaerts, D., Peirs, J., Van Brussel, H.: A mechatronic approach

to microsystem design. IEEE/ASME Trans. Mechatronics 3(1),

24–33 (1998)

10. Skantze, K.: Method and device for secure wireless transmission

of information. US Patent 7,278,017, 2 Oct 2007

11. Euchner, J.A., Coffy, J.H., Obrea, A.: System and method for

annotating documents. US Patent 7,111,230, 19 Sept 2006

12. www.fruits-it.com/request.php?8. Accessed 20 Dec 2007

13. http://www.otmtech.com. Accessed 6 Dec 2007

14. Clijnen, J., Reynaerts, D., Van Brussel, H.: Design of an optical

tri-axial force sensor. In: Bar-Cohen, Y. (ed.) Proceedings of

SPIE, vol. 4946, Transducing Materials and Devices, March

2003, pp. 129–136

15. Shimizu, H., Kiyono, S., Motoki, T., Gao, W.: An electrical pen

for signature veriﬁcation using a two-dimensional optical angle

sensor. Sensor Actuator 111, 216–221 (2004)

Dimensionality Reduction

The process of reducing the number of features used by

a classiﬁer in order to decrease measurement cost,

increase classiﬁcation accuracy, and mitigate the pro-

blems associated with the curse-of-dimensionality. Fea-

ture selection or feature extraction techniques is used to

deduce an optimal (or, from a practical standpoint,

sub-optimal) set of features from a pool of available

features. Common examples include sequential for-

ward selection (SFS), sequential backward selectio n

(SBS), sequential forward ﬂoating search (SFFS), etc.

▶ Fingerprint Sample Synthesis

▶ Fusion, Feature-Level

Diode

Diode is an electronic component that allows the pas-

sage of current in only one direction. A light emitting

diode (LED) is an electronic semiconductor diode that

emits a single wavelength of light when electric current

passes through it. A photodiode is a semiconductor

diode that allows current to ﬂow when it absorbs

photons (light).

▶ Digitizing Table t

Diphones

A diphone brackets exactly one phoneme-to-phoneme

transition. Diphone boundaries are usually positioned

228

Dimensionality Reduction

near the midpoint of the most stationary (non-

changing) region of two consecutive phonemes.

Theoretically, a phoneme inventory of 50 could give

rise to up to 2,500 diphones, but not all diphones exist

in a given language. For general American English, a

minimum of about 1,500 diphones are necessary.

▶ Voice Sample Synthesis

Disclosure Check

▶ Background Checks

Discriminative Classifier

A discriminative classiﬁer is a classiﬁcation algorithm

than learns a borer; one side it labels one class, the

other side it labels another. The border is chosen to

minimize error rate, or some correlated measure, ef-

fectively discriminating between the classes.

▶ Fusion, Quality-Based

Dissimilarity

▶ Palmprint Matching

Distortion

Variances in the ridges and features of a ﬁngerprint caused

by deposition pressure or movement are distortion.

▶ Universal Latent Workstation

Distributed Computing

Distributed computing refers to the paradigm of not

having a central computing node in a sensor network.

Distributed computing uses parallel computations

over a multiple processing units, connected by a com-

munications network. Distributed computing allev i-

ates the need for having an extremely powerful

central computing node in a network, sacriﬁcing per-

formance to obtain robustness to partial failure of the

network both in terms of individual node failures as

well as failure on the communication links.

▶ Surveillance

Distributed Detection

▶ Fusion, Decision-Level

Distributed Inference Making

▶ Fusion, Decision-Level

DNA Analysis

▶ Forensic DNA Evidence

DNA Fingerprinting/DNA Profiling

DNA ﬁngerprinting is a term coined by Sir Alec Jeffreys

describing the multi-locus probes results obtained in

DNA Fingerprinting/DNA Profiling

229