Scalable Imaging Device using Line Scan Camera for Use in

Biometric Recognition and Medical Imaging

Michal Dvořák

, Ondřej Kanich

and Martin Drahanský

Faculty of Information Technology, Brno University of Technology, Brno, Czech Republic

Keywords: Imaging System, Fingerprint, Hand Geometry, Line-Scan Camera, Skin Diseases.

Abstract: In this paper, a novel imaging system for use in a biometric or medical application utilizing a line scan camera

is being presented. The system utilizes a linear motion system to achieve a variable field of view as per

operators' demands, while maintaining the high resolution per unit area. Use cases are presented using several

demonstrations of possible applications. Fingerprint quality evaluation algorithms are showing applicability

as a biometric-enabled system. Dermatological application is demonstrated by using the acquired images to

perform a measurement of common nevi. Further uses in wound treatment and other biometrics such as hand

geometry recognition and palmprint recognition are discussed.

1 INTRODUCTION

Whether the goal is to perform medical imaging for

dermatology or wound care purposes, or perhaps data

for biometric applications are required, there is a need

to collect the images of a human body in a precise,

repeatable, and high-quality manner. Despite cameras

being widespread, and their parameters soaring in the

last couple of years, when the application demands

highly detailed images of large areas of the human

body, it can be quickly discovered that even the

available high-resolution cameras are insufficient. It

is then necessary to either come up with a way to

utilize the less detailed imagery or to find a way to

generate a sufficient number of high-quality 2D sub-

images that can be reconstructed into the final image.

Both methods have their drawbacks. What if,

however, instead of using a traditional 2D camera

sensor with a fixed number of pixels per area, a sensor

with variable field of view was made?

The goal of this work is to present a design and a

prototype of an optical scanning device, utilizing a

line scanner instead of a traditional 2D sensor. A

device that can change the field of view based on the

operators' needs and thus allowing for scanning body

parts of various sizes without having to change the

https://orcid.org/0000-0003-3265-6955

https://orcid.org/0000-0003-0093-8536

https://orcid.org/0000-0002-9321-7385

physical configuration of the device. A device that

can facilitate scanning of a finger, a hand or a whole

arm without having to sacrifice an image quality.

In this paper, a possible configuration of the

device is demonstrated by capturing images of human

hands and acquiring a small database. This decision

has been made to allow the use of biometric quality

evaluation to establish the performance of the device.

1.1 Use Cases

The intended application of this device is a medical

imagining in the visible, near-infrared or ultraviolet

spectrum with a focus on usability in dermatology

and potentially a wound treatment such as shown in

(Dick et al., 2019), (Zhang et al., 2017), (Deng et al.,

2017) and the technologies developed for this

purpose, most notably by (Korotkov et al., 2015) and

(Haeghen et al., 2000). The device was designed for

tracking the change or rate of change of selected

features such as various skin lesions and scar tissues.

1.2 Operation Principle

Unlike the traditional 2D CMOS or CCD sensor,

where during the acquisition the whole target object

is projected onto the sensor and is downloaded from

160

Dvo

rák, M., Kanich, O. and Drahanský, M.

Scalable Imaging Device using Line Scan Camera for Use in Biometric Recognition and Medical Imaging.

DOI: 10.5220/0010342601600168

In Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 1: BIODEVICES, pages 160-168

ISBN: 978-989-758-490-9

Figure 1: Line scanner principle (Fermum, 2016).

the camera after that, the proposed device utilizes a

line scanner instead.

The general operating principle of line scanner

can be seen in Figure 1. During the acquisition, the

camera captures information regarding a line whose

width w

is determined by the camera and camera

lens parameters. In this paper, the axis that

corresponds to the acquired line will be described as

a sensor axis and the perpendicular axis will be called

the movement axis.

To reconstruct the complete image of the target,

either the camera or the target needs to move along

the movement axis at a predefined speed v

and the

acquisition must be performed at a rate of f.

𝑓

= 1/𝜏

(1)

𝜏= 1 2

⁄

∙𝑤



𝑣



⁄

(2)

Where f [Hz] is the rate of acquisition in lines per

second. 𝜏 [s] is the period between acquisitions.

[mm] is the resolvable width which is required and

[mms

-1

] is the speed of the system. The factor of ½

is necessary to meet the Nyquist–Shannon (NS)

sampling theorem (Nyquist, 1928) due to the

movement of the system.

1.3 Optical Principle

Given the distance to the target object and the size of

the chosen sensor, the camera lens needs to be

calculated. To determine the appropriate camera lens,

the required field of view fov

needs to be considered.

In this case, fov

describes the maximum dimension

(width) of the target that can be observed with chosen

optics.

𝑓𝑜𝑣



represents a field of view expressed as an

angle.

𝑓

𝑜𝑣



=2𝑠∙tan (𝜃/2)

(3)

𝑓

𝑜𝑣



= 2 ∙ arctan (h/2f)

(4)

s is a distance to our target, f is the focal length of

our chosen lens and h is the actual size of the sensor.

The focal distance (Ray, 2004) f can be calculated

using the required distance to the object, also known

as object distance g, and ratio of sensor size to target

size, also known as the magnification m (Ray, 2004).

𝑓

=𝑚𝑔 (1+𝑚)

⁄

(5)

The size of the smallest resolvable object 𝑤



which can be scanned in the sensor axis can be

determined from the resolution of used sensor r

and

the

𝑓𝑜𝑣

using the following formula.

𝑤



=2∙

𝑓

𝑜𝑣



/𝑟



(6)

Where the factor of two ensures NS sampling

theorem is met. For completeness, the following

formula derived from (2) and (3) denotes the

resolvability limit in the axis of the movement.

𝑤



=2∙𝑣



𝑓

⁄

(7)

For the traditional 2D sensor, the field of view in

both dimensions would need to be considered,

however, for line scanner, the field of view of the

movement axis is defined purely by the physical

dimension of the scanner in the axis of the movement.

For finite image, the field of view of the image in

the axis of the movement

𝑓𝑜𝑣



can be simply

defined as a function of time t.

𝑓

𝑜𝑣



=𝑣



∙𝑡

(8)

1.4 Requirements for Biomedical and

Biometric Applications

The most known biometric systems are touch-based

fingerprint sensors used in mobile phones or laptops.

Other devices use hand geometry, finger vein or palm

vein for the recognition process (Debiasi et al., 2018).

One thing these devices often have in common is the

necessity to touch the sensing area. Especially now,

the hygienic demands dictate that the devices used by

multiple people need to be cleaned and disinfected at

regular intervals. This paper offers an approach where

no optical element needs to be touched.

The human hand contains much information that

can be used for biometric recognition (Drahanský &

Kanich, 2019). The focus of this article is to analyze

fingerprint and hand geometry characteristics with

limited discussion concerning the feasibility of

palmprint and vein extraction.

The spatial resolution used for fingerprint

recognition is typically 500 dpi (dots per inch) for the

commercial devices but can be performed with

images with resolution as low as 300 dpi (Drahanský

et al., 2018).

Customarily, the hand is placed onto a designated

surface to limit the user’s movement. High contrast

Scalable Imaging Device using Line Scan Camera for Use in Biometric Recognition and Medical Imaging

161

images are desirable for valleys and ridges to be

easily distinguishable.

For the hand geometry, the resolution

requirements are lower than for fingerprints. For

successful hand shape extraction, the scanned area

needs to be large enough for the whole hand up to the

wrist to be visible.

For biomedical purposes, the device needs to be

calibrated to give the device an ability to measure the

size of the studied objects. The external light source

and repeatable scanning allows the comparison of the

studied objects and evaluate the rate of change in size,

colour or structure (Dick et al., 2019).

2 DEVICE DESIGN

In this part, the general requirements and consequent

design of the scanner will be discussed along with a

list of components that have been used for the

prototype described in this paper.

2.1 Optical Design

Following parameters are defined.

1. The spatial resolution of the image is to be at

least 350 dpi.

2. The scanned area is to be at least

400 × 180 mm.

The first requirement allows to resolve objects of

approximately 0.145 mm. It ensures that both the

biometric applications and the calibrated

measurement for a medical imagining application can

be performed. Both requirements are used to define

linear scanner itself, as the minimum resolution can

be derived from them in the following manner.

𝑅_𝑚𝑖𝑛 = 350 ∙ (400/25.4)  5,512

(9)

Based on this requirement, a line scanner

raL6144-16gm Basler with the resolution of 6,144 px

has been chosen as suitable.

Formula (5) can be used to find a suitable camera

lens. The object distance of the device is 535 mm, as

further discussed in subsection 2.2. Given the

dimensions of the chosen sensor and the target area

dimensions, the magnification needs to be at least

0.096 to guarantee the 350 dpi requirement. Using the

formula (5), the focal length can be calculated to be

47 mm. 50 mm camera lens has been used due to its

availability. This guarantees a resolution of 350 dpi

for the area of 417.1 mm. Camera lense AF Nikkor

50mm f/1.8D has been used in the device. The

Chromasens Corona II type C with LED-control unit

XLC4-1 has been used for illumination.

2.2 Mechanical and Electrical Design

The main computing unit is a single board computer

(SBC) Raspberry Pi 3. It has been decided that the

movement subsystem will be controlled by a separate

microcontroller (MCU) Arduino Uno.

Figure 2 illustrates how are the individual

components connected. The functionality and

purposes of individual subsystems are further

described in subsection 2.3.

Figure 2: Functional diagram of the device, illustrating

individual subsystem and communication interfaces.

For power, the device utilized three Manson NP-

9615 DC regulated power supplies to generate 48 V,

24 V, 12 V and 5 V for voltage branches. The

summary of electrical requirements for the used

components can be found in Table 1.

The conversion of rotational movement of the

stepper motor into the linear motion necessary for the

scanning has been realized using a linear motion

system with a threaded rod NL208TR-1200 made by

T.E.A. Technik (T.E.A Technik s.r.o., 2020). This

system, considering the method of scanning, allows

for objects up to 120 × 43 cm in size to be scanned

while maintaining the 350 dpi spatial resolution or in

other words, it functions as approximately 100 Mpx

industrial-grade camera.

BIODEVICES 2021 - 14th International Conference on Biomedical Electronics and Devices

162

Table 1: Electrical parameters of components.

Component Voltage [V]

Maximum

current [A]

raL6144-16gm 12 0.375

XLC4-1 24 3

CORONA II 24 1.8

EM705 48 -

86HS45 48 4.2

RPI3 5 2.5

Arduino UNO r3 12 -

The Figure 3 represents a front view of the

hardware setup in schematics form, along with the

physical dimensions in mm. The dimension 345 mm

represents the maximum range the scanner can move

and, therefore, also the maximum dimension of the

resulting image along this axis. During the database

collection, the users approached the device from the

front, placed the hand onto the Hand placement

platform and the scanner performed a complete image

acquisition by moving from left to right.

Figure 3: Construction design of the scanner.

Figure 4: The function calls sequence during the scanning

process.

2.3 Control System

The SBC either directly or indirectly controlled each

part of the device. Figure 4 shows the control

structure and the way in which the modules are

sequentially called.

2.3.1 Light Control

The SBC was communicating with the XLC4-1 unit

using Telnet interface over the ethernet. Series of

simple commands provided by the unit's

documentation were used to set the intensity of the

light and to turn the light module on and off.

Commands to monitor the status of the unit and the

module were also utilized.

2.3.2 Movement Control

The MCU has performed movement control to avoid

unexpected interruptions of the SBC. Using the

USART interface, a command from SBC is sent to

MCU that contains a piece of information about

requested direction and distance to be travelled. MCU

parses this command and using the HW timer and

counter, it generates a pulse sequence with a

frequency of 62.5 kHz for the appropriate number of

ticks.

EM705 stepper driver is set for micro-stepping at

12,800 steps per revolution. This, combined with the

physical property of NL208TR-1200 which performs

a lateral movement of 4 mm per one revolution, gives

us a precise function that defines the distance

travelled d in millimetres as a function of time t in

seconds in the following manner.

𝑑(𝑡) = 4 ∙ (62,500 ∙ 𝑡) 12,800

⁄

(10)

The prototype for the database collection used a

scanning time of 10,000 ms (10 s), covering

195.31 mm of distance.

2.3.3 Line Scanner Control

The line scanner is controlled by SBC using the GigE

interface and Basler API. The script developed for

this purpose handles proper initialization, data

grabbing and image construction. All triggering and

frame grabbing is performed using software scripts.

The prototype due to standard dimensions of the

target performs acquisition of 5,120 lines at a rate of

500 lines per second. Considering the movement

speed of the linear movement system, the chosen rates

cause a line width to be 0.039 mm. Due to the

apparent difference of the vertical pixel size from the

Scalable Imaging Device using Line Scan Camera for Use in Biometric Recognition and Medical Imaging

163

horizontal pixel size, the image is upscaled by a factor

of two in the x-axis prior being saved.

3 DATA COLLECTION

The choice has been made to use a visible light source

for image acquisition. For the application, it was

chosen to test fingerprint recognition which requires

very precise images of the ridges and recognition

based on hand geometry which can use both sides of

the hand.

3.1 Database Collection

Per the design decisions outlined in previous sections

of this paper, the device generates images with the

resolution of 6,144 × 5,120 px that are upscaled to

12,288 × 5,120 px for easier visual inspection. The

scanned area is 417.1 × 195.3 mm which corresponds

to the spatial resolution of 375 dpi in the scanner axis

and 666 dpi in the movement axis. In Figure 5, two

sample images are presented.

Figure 5: Example of dorsal hand surface image (left) and

palmar surface image (right).

The image collection was done without adhering

to any specific condition, in an ordinary room with

daylight and artificial light present. Volunteers came

into the room and one or both of their hands were

acquired from palmar and/or dorsal side.

Overall, 145 images were acquired. These

contained 77 hands from the volunteers in the 20-30

age group, mixed gender. 56 of these images are from

the palmar side, remaining 89 from the dorsal side.

3.2 Images Applications

It is essential to remind the reader that the device was

developed primarily as a scanning method and that

the biometric quality evaluation serves primarily to

gauge the image quality.

3.2.1 Biometric Recognition: Fingerprints

Due to large scanned area, the preprocessing aims to

localize and extract fingerprints, these subimages are

then enhanced as the algorithms for the fingerprint

evaluation, generally expecting binarized or at least

normalized data.

Hand extraction started with coarse image

alteration. In this step, a rectangle of 7,800 px to

4,300 px is isolated from position x = 4,500 px and y

= 225 px. Cropped images contain only the hand and

a part of the wrist. Images were rotated so that fingers

pointed upwards and background was thresholded

away.

Using convolution-based localization algorithms,

the probable position of fingers was determined.

Figure 6 shows result of the finger localization. After

getting the finger images, histogram equalization and

local thresholding methods are used to normalize the

image and to increase the contrast of the ridges.

Result can be seen in the Figure 7.

Figure 6: Input image (left) and isolated fingers (right).

Figure 7: Example of two processed fingerprints.

280 images containing 230 fingerprints were

acquired from the images. The fingerprints were then

evaluated using VeriFinger (version 10)

(Neurotechnology, 2018) and FiQiVi (Dejmal, 2019).

BIODEVICES 2021 - 14th International Conference on Biomedical Electronics and Devices

164

The median VeriFinger quality was 32 (maximum 60,

minimum 10), FiQiVi showed a median of 38

(maximum 57 and minimum 19). For biometric

recognition the VeriFinger developer

Neurotechnology recommends the quality of at least

40 for identification and 30 for verification. This has

also been experimentally verified by (Alsmirat et al.,

2018).

While the quality of the fingerprints may seem

low, it is an impressive score. Firstly, the device to

facilitate the large area acquisition used a 375 dpi

resolution, which is lower than it is usual for

fingerprint sensors. Secondly, unlike the almost

absolute majority of sensors, this device does not use

a scanning surface against which the finger is to be

pressed. Thus, it reduces the available scanning area

due to the finger geometry, as it is not pressed against

a glass surface, as well as decreases the contrast due

to the absence of total internal reflection. This

concept can be seen in (Maltoni et al., 2009) Thirdly,

this evaluation contained also thumbs which were

usually acquired at an angle. Lastly, only basic image

pre-processing was done with no fingerprint

enhancement algorithms being used. Therefore, the

fact that despite all the disadvantages, the device still

manages to meet the fingerprint biometry baseline is

a relatively strong proof of its viability.

3.2.2 Biomedical Applications

The device is mainly intended to be used for

monitoring hand diseases and disorders, such as

eczema, as well as various skin growths, such as

nevus, to monitor their size and area. In the previous

section, it was shown that the device is able to

distinguish ridges that have a width from 0.2 to

0.5 mm (Drahanský et al., 2011). Based on fixed

optics and calibrated system, precise measurements

can be performed.

Figure 8 (top) provides a detailed image of nevi

on one user’s hand. Due to the system being

calibrated, we can determine that a bounding box of

34 × 29 px and 24 × 19 px that can be used to

surround the nevi corresponds to size of

1.23 × 1.13 mm and 0.87 × 0.74 mm respectively.

The analysis of the size of the nevi can be

immediately performed as well as saved and

compared during subsequent scanning for any

change. The same process can be used to observe the

rate or any changes in the wound healing process.

This can be seen in Figure 8 (bottom) were mutilated

Figure 8: Detail of nevi (top). Regenerated friction ridges

on the fingertip (bottom).

finger is present but the current images clearly show

that the friction ridges are restored on the remainder

of the finger.

3.2.3 Biometric Recognition: Hand

Geometry

The images collected by the scanner are of sufficient

quality to perform recognition based on hand

geometry if other biometrics are unavailable, for

example, due to injuries such as a burn or skin

diseases.

As the acquisition process is calibrated and the

images can be used to perform measurements of the

target object, absolute measurements may be used as

a feature set for the recognition algorithms, such as

described by (Pititeeraphab & Pintavirooj, 2018) and

(Wirayuda et al, 2013) or relative measurement, as

described by (Siswanto et al, 2013).

Majority of algorithms perform segmentation and

work with a binarized image of the hand, which these

images can also be converted to. The feature set can

then be constructed from such features as outlined in

the Figure 9. Given the knowledge that pixel's height

corresponds to 0.039 mm and the pixel's width

corresponds to 0.036 mm. Using this information,

any selected length of the image may be recalculated

into a real-world dimension.

Figure 9: Proposed hand geometry features.

Scalable Imaging Device using Line Scan Camera for Use in Biometric Recognition and Medical Imaging

165

3.2.4 Biometric Recognition: Other

Characteristics

Due to the image size and resolution, additional

biometric features can be extracted. Figure 10 shows

the enhanced image of the palmar side of the hand

indicating a multitude of data for possible biometric

(e.g. palmprint recognition). In this paper, visible

light source has been used, however, should the other

wavelength be used, other data may be acquired. The

use of near-infra lighting near the 850 nm wavelength

would allow for an observation of veins such as

(Huang et al., 2018), again usable for biometric or

medical purposes.

Figure 10: Example of enhanced image of the palmprint.

3.3 Discussion of the Results

The presented results and images serve as a

demonstration of the viability of line scan cameras for

biomedical and biometric application. The ability to

perform scans of various lengths gives this system

flexibility the devices with traditional 2D sensors

lack. It can also achieve very good precision on the

whole scanned area. There are some disadvantages as

well. Scanned user has to remain calm during the

scanning process and line cameras can be expensive.

3.3.1 The Device as a Biometric System

In the article, fingerprint and hand geometry

recognition has been discussed and its efficiency

evaluated. It can be concluded that while the device

would need to be modified for it to be a more effective

biometric system, the values from commercial and

academic quality assessment algorithms indicate that

even as it is, the device could be used for biometric

application. This, more than anything, speaks of the

image quality received from the device and proves its

efficacy in scanning live objects which, to the

knowledge of the authors, has not been widely

explored for line scan cameras.

Applicability of the hand geometry recognition

and palmprint recognition has been discussed, while

the usability of the palmprint is implied by the fact

that the images may be used for fingerprint

recognition. There is, to the knowledge of the authors,

no available tool to measure the image quality for the

hand geometry recognition. Therefore, only relevant

research into existing literature has been done.

3.3.2 The Device as a Biomedical System

Based on the available literature, the skin images of

sufficient quality may be used to analyze various skin

growths. It has been demonstrated, both in the design

of the device and in the final images, that the skin

images offer sufficient quality to perform imaging

and extract a precise measurement of the various skin

objects. Should the scanning be repeated, a change in

time can be observed.

4 CONCLUSIONS

This paper introduces an optical scanning device

using line scan camera, linear movement system, light

unit, single board computer and microcontroller

capable of producing images of the variable field of

view without having to make concessions to the

quality due to its mechanical properties.

The laboratory prototype of the device was

constructed and pilot data collection consisting of 145

human hands, of which 56 was palmar side up, was

conducted. The function and the performance of the

device have been tested to determine its efficiency as

a biometric and biomedical system. The first test was

focused on the possibility of using acquired images

for fingerprint recognition. Overall, 280 fingerprint

images were extracted from the database and

analyzed using two fingerprint quality assessment

systems. The image quality of the majority of the

extracted fingerprints has been determined to be

sufficient for fingerprint verification and

identification.

Dorsal and palmar images of hand were examined

for the usage in hand geometry recognition as well as

performing measurements of various skin features

usable for biomedical application. These

measurements are suitable for monitoring nevi, skin

diseases analysis and a rate of change in the wound

healing process. Based on the analysis, it was decided

that the detail of the images is sufficient.

BIODEVICES 2021 - 14th International Conference on Biomedical Electronics and Devices

166

To sum it up, the device can obtain high-quality

images. The examined object did not have to touch

the sensing area. Size of the images is limited by the

defined width of the camera, but scalable in the

movement axis, limited only by the linear motion

system's properties. Within the limits of the motion

system, the change of the image size due to the

movement axis does not require any physical changes

to the device and can be changed programmatically

by the operator.

For further research, the device could be enlarged

and used as a full-frontal body scanner. Optimization

of the biometric extraction process would lead to

higher performance rate for the biometric applications.

ACKNOWLEDGEMENTS

This research has been realized under the support of

the following grants: Technology Agency of the

Czech Republic from the ÉTA programme "Survey

and education of citizens of the Czech Republic in the

field of biometrics – TL02000134", "Reliable, Secure,

and Efficient Computer Systems" – internal Brno

University of Technology project FIT-S-20-6427.

REFERENCES

Alsmirat, M. A., Al-Alem, F., Al-Ayyoub, M., Jararweh,

Y., Gupta, B., 2018. Impact of digital fingerprint image

quality on the fingerprint recognition accuracy.

Multimedia Tools and Applications, 78(3), 3649-3688.

https://doi.org/10.1007/s11042-017-5537-5

Debiasi, L., Kauba, C., Prommegger, B., Uhl, A., 2018.

Near-Infrared Illumination Add-On for Mobile Hand-

Vein Acquisition. 2018 IEEE 9th International

Conference on Biometrics Theory, Applications and

Systems (BTAS). https://doi.org/10.1109/BTAS.

2018.8698575

Dejmal, D., 2019. Analýza systémů pro měření kvality

otisku prstů [Bachelor’s thesis, Brno University of

Technology]. https://www.fit.vut.cz/study/thesis-

file/19344/19344.pdf

Deng, Z., Fan, H., Xie, F., Cui, Y., Liu, J., 2017.

Segmentation of dermoscopy images based on fully

convolutional neural network. 2017 IEEE International

Conference on Image Processing (ICIP).

https://doi.org/10.1109/icip.2017.8296578

Dick, V., Sinz, C., Mittlböck, M., Kittler, H., Tschandl, P.,

2019. Accuracy of Computer-Aided Diagnosis of

Melanoma. JAMA Dermatology, 155(11), 1291.

https://doi.org/10.1001/jamadermatol.2019.1375

Drahanský, M. (Ed.), 2018. Hand-Based Biometrics:

Methods and technology. IET, ISBN 978-1-78561-224-

Drahanský, M., Kanich, O., 2019. Influence of Skin

Diseases on Fingerprints. In Nait-Ali, A. (Ed.),

Biometrics under Biomedical Considerations. Springer.

https://doi.org/10.1007/978-981-13-1144-4

Drahanský, M., Orság, F., Dvořák, R., Hájek, J., Váňa, J.,

Herman, D., Kněžík, J., Marvan, A., Lodrová, D.,

Doležel, M., Hanáček, P., Mráček, Š., Stružka, J., 2011.

Biometrie. Cumpter Press s.r.o. ISBN 978-80-254-

8979-6.

Fermum, L., 2016 (n.d.). Line scan camera basics.

Retrieved November 05, 2020, from https://www.

vision-doctor.com/en/line-scan-cameras/line-scan-

camera-basics.html

Haeghen, Y., Naeyaert, J., Lemahieu, I., Philips, W., 2000.

An imaging system with calibrated color image

acquisition for use in dermatology. IEEE Transactions

on Medical Imaging, 19(7), 722-730. https://doi.org/

10.1109/42.875195

Huang, Q., Hu, K., Zhou, P., Luo, Y., Wu, L., 2018. Design

of Finger Vein Capturing Device Based on ARM and

CMOS Array. 2018 2nd IEEE Advanced Information

Management, Communicates, Electronic and

Automation Control Conference (IMCEC).

https://doi.org/10.1109/imcec.2018.8469403

Korotkov, K., Quintana, J., Puig, S., Malvehy, J., Garcia,

R., 2015. A New Total Body Scanning System for

Automatic Change Detection in Multiple Pigmented

Skin Lesions. IEEE Transactions on Medical Imaging,

34(1), 317-338. https://doi.org/10.1109/tmi.2014.235

7715

Maltoni, D., Maio, D., Jain, A. K., Prabhakar, S., 2009.

Handbook of Fingerprint Recognition. Springer,

Second Edition, ISBN 978-1-84882-253-5.

Neurotechnology, 2018. MegaMatcher 10.0, VeriFinger

10.0, VeriLook 10.0, VeriEye 10.0 and VeriSpeak 10.0

SDK – Developer’s Guide. Neurotechnology, Version:

10.0.0.0.

Nyquist, H., 1928. Certain Topics in Telegraph

Transmission Theory. Transactions of the American

Institute of Electrical Engineers, 47(2), 617-644.

https://doi.org/10.1109/t-aiee.1928.5055024

Pititeeraphab, Y., Pintavirooj, C., 2018. Identity

Verification Using Geometry of Human hands. 2018

11th Biomedical Engineering International Conference

(BMEiCON).

https://doi.org/10.1109/bmeicon.2018.8609986

Ray, S. F., 2004. Applied photographic optics: Lenses and

optical systems for photography, film, video, and

electronic imaging. Oxford: Focal Press, Third edition.

ISBN 978-0-24051-540-3.

Siswanto, A., Tarigan, P., Fahmi, F., 2013. Design of

contactless hand biometric system with relative

geometric parameters. 2013 3rd International

Conference on Instrumentation, Communications,

Information Technology and Biomedical Engineering

(ICICI-BME). https://doi.org/10.1109/icici-bme.2013.

6698492

T.E.A. TECHNIK s.r.o., 2020. Lineární osa se šroubem

NL. T.E.A. TECHNIK s.r.o. Retrieved November 04,

Scalable Imaging Device using Line Scan Camera for Use in Biometric Recognition and Medical Imaging

167

2020, from https://www.teatechnik.cz/linearni-osa-

sroubem-nl/

Wirayuda, T. A., Kuswanto, D. H., Adhi, H. A., Dayawati,

R. N., 2013. Implementation of feature extraction based

hand geometry in biometric identification system. 2013

International Conference of Information and

Communication Technology (ICoICT).

https://doi.org/10.1109/icoict.2013.6574583

Zhang, X., Wang, S., Liu, J., Tao, C., 2017. Computer-

aided diagnosis of four common cutaneous diseases

using deep learning algorithm. 2017 IEEE International

Conference on Bioinformatics and Biomedicine

(BIBM). https://doi.org/10.1109/bibm.2017.8217850

BIODEVICES 2021 - 14th International Conference on Biomedical Electronics and Devices

168