Multimodal Biometric Recognition Systems based on Physiological

Traits: A Systematic Mapping Study

Hind Es-Sobbahi

, Mohamed Radouane

and Khalid Nafil

Software Project Management Research Team, ENSIAS, Mohammed V University, Rabat, Morocco

Keywords: Multimodal, Physiological, Biometric, Recognition, Image Processing, Systematic Mapping Study,

Systematic Literature Review, Deep CNN Methods.

Abstract: Context: Biometric systems are fundamental to protect against identity theft and illegitimate access. However

most of them are unimodal and have several drawbacks such as: noisy data, intra-class variation, inter-class

similarity, non-universality and spoofing attacks. Hence, multimodal biometric recognition systems (MBRS)

are increasingly in demand to overcome these limitations.

Objective: This work aims to aggregate and synthesize available studies and provide a historical and

geographical classification in order to guide researchers in their choices of biometric traits (BT) combinations

and image processing (IP) techniques. Therefore, we conducted a Systematic Mapping Study (SMS). Method:

We analysed 247 relevant articles to answer the research questions according to inclusion and exclusion

criteria, namely: country, source and year of publication, BT combinations and IP techniques employed.

Results: According to our results, India tops the list; iris, fingerprint and face are the most requested by

researchers. Concerning IP techniques used, PCA Algorithm leads (24%), followed equally (14%) by LBP

and Deep CNN.

Conclusion: This SMS was produced to guide stakeholders in choosing the most relevant configuration

between of BT and IP methods when designing an MBRS. Findings are interesting as they provide a detailed

overview of aspects that can impact the performance of a system.

1 INTRODUCTION

The pressing need to use computer security

applications has increased considerably over time,

making automatic personal authentication of great

importance. Biometrics refers to the measurement and

statistical analysis related to human characteristics.

The main advantage of biometric authentication is

that each person can be identified with a high degree

of accuracy based on intrinsic physical or behavioral

characteristics (Deriche, 1998). It has long been

defined as a vigorous method of authenticating

people. With new technological advances, biometric

recognition systems have become an emerging

solution to

solve the problems related to revealing

person's identity. Unimodal biometrics systems face

various problems such as intra-class variations, non-

universality, noisy data, restricted degrees of freedom,

https://orcid.org/0000-0003-2562-0647

https://orcid.org/0000-0003-2898-889X

https://orcid.org/0000-0003-4963-0225

unacceptable error rates and spoof attacks (Ross &

Jain, 2004). Therefore, the orientation towards a

multimodal biometric system, which is based on

multiple sources of information, is an alternative to

solve these problems and improve the performance.

Multimodal biometric systems are more reliable due

to the presence of multiple proofs of identity and they

can reduce vulnerability against spoofing attacks by

using, at the same time, different modalities and

varying biometric sources of information. In general,

they provide better recognition performance than

systems based on a single biometric modality (Jain,

Nandakumar & Ross, 2005).

The studies defined in this work show that

researchers are constantly proposing new models of

biometric systems to achieve maximum authenticity,

accuracy and reliability. (Nguyen, Fookes, Jillela,

Sridharan, & Ross, 2017) review the state of the art

556

Es-Sobbahi, H., Radouane, M. and Naﬁl, K.

Multimodal Biometric Recognition Systems based on Physiological Traits: A Systematic Mapping Study.

DOI: 10.5220/0012128600003538

In Proceedings of the 18th International Conference on Software Technologies (ICSOFT 2023), pages 556-563

ISBN: 978-989-758-665-1; ISSN: 2184-2833

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

design and implementation of iris-recognition-at-a-

distance (IAAD) systems, present existing IAAD

systems and also discusse current research challenges

while providing recommendations for future research

in IAAD. (Shaheed, Liu, Yang, Qureshi, Gou, & Yin,

2018) present a detailed review of finger vein

recognition algorithms. The comparative studies

indicate that the accuracy of finger vein identification

methods is up to the mark. (Shaheed, Mao, Qureshi,

Kumar, Abbas, Ullah, & Zhang, 2021)

summarize and

investigate various traditional and deep learning based

biometric modalities. An in-depth examination of the

biometric steps of several modalities using different

levels such as pre-processing, feature extraction and

classification is presented in detail. The result of the

comparison indicates that there is still a need to

develop a robust physiology-based method to advance

and improve the performance. (Bharadwaj, Vatsa &

Singh, 2014) introduce extensive reviews of biometric

technology. (Unar, Seng & Abbasi, 2014)

present the

characteristics, strengths and limitations of existing

techniques for assessing the quality of various

biometric traits, including fingerprints, iris and face...

ECG and Lip Print are two emerging biometric

modalities. ECG can be combined with other robust

biometrics such as fingerprints and iris to provide a

reliable multimodal biometric system. The Lip Print

can be combined with face and voice to develop a

user-friendly biometric system (Chauhan, Arora &

Kaul, 2010). (Rui & Yan, 2018) classify the existing

biometric authentication systems by focusing on the

security and privacy solutions. They figure and

specify a number of important research directions that

deserve special efforts in future research. Finally,

(Abo-Zahhad, Ahmed, & Abbas, 2014) aim to review

previous studies related to the use of ECG and PCG

signals in human recognition and discusse the most

important techniques and methodologies used by

researchers in the preprocessing, feature extraction

and classification of the ECG and PCG signals.

Based on this and according to the knowledge of

the author, no mapping study has been carried out to

date with a focus on multimodal biometric recognition

systems based on physiological traits. Therefore, we

were interested by conducting this study to diagnose,

analyze and summarize published articles in IEEE

Xplore, Science Direct, Springer, Wiley, ACM,

Scopus and Web of Science databases related to this

type of systems. During the period between 2010 and

2022, we have selected 1289 papers. After an

exhaustive selection study we have identified 247

relevant papers.

The remainder of this paper is organized as

follows: Section 2 describes the research

methodology used to achieve this work. Section 3

presents and discusses the mapping results. Section 4

discusses the results obtained and presents the

implication for researchers. Finally, conclusion is

presented in section 5.

2 MAPPING PROCESS

Mapping studies can be very useful to researchers in

establishing a solid base for further research in a

defined area. They are based on the same

methodology applied in the development of SLRs,

except that the difference lies in the objective of each

one of them.

A systematic mapping study helps to structure the

type of research reports and findings that have been

published by organizing them and often provides a

visual summary, the map, of its findings. It often

requires less effort while providing a big overview

(Petersen, Feldt, Mujtaba & Mattsson, 2008).

To perform this study, we follow the mapping

process suggested by (Petersen, Feldt, Mujtaba &

Mattsson, 2008), steps are outlined in figure 1, each

step of the process is described in detail in the

following subsections with the results they provide.

2.1 Definition of Mapping Questions

2.1.1 Mapping Questions (MQs)

According to (Petersen, Vakkalanka & Kuzniarz,

2015), the research questions in mapping studies are

general as they aim to uncover research trends. Thus,

we present this mapping study in order to provide a

comprehensive overview of physiological-based

biometric multimodal recognition systems and to

identify the amount and the type of published

researches between 2010 and 2022. The following

mapping questions were identified:

 MQ1: How publications are distributed across

countries?

Motivation: To identify the universities and research

laboratories working on this subject around the world.

 MQ2: What is the quality of relevant articles

according to conference and/or journal rank?

Motivation: To know the degree of scientific

relevance of the different contributions.

 MQ3: According to MQ1 and MQ2, is there a

correlation between the quality of the selected articles

and the scientific production?

Motivation: To know if there is a proportional

relationship between the quality and the quantity of

Multimodal Biometric Recognition Systems based on Physiological Traits: A Systematic Mapping Study

557

the published works according to conference and/or

journal rank.

 MQ4: What is the annual trend of publications?

Motivation: To give researchers the opportunity to

position themselves well in relation to this

disciplinary field.

 MQ5: Where were the selected studies

published?

Motivation: To know if there is a cumulative and

progressive interest from the scientific community

over time.

 MQ6: What are the most used biometric traits

combinations studied in the selected papers?

Motivation: To find the most requested combinations

of biometric traits in the studied systems.

 MQ7: What are the main image processing

techniques identified in the selected papers?

Motivation: To know the different image processing

techniques used when designing systems and be able

to classify them according to the recorded

performance.

2.2 Search Strategy

To perform this study, it was imperative to answer the

predefined research questions. For this we have

adopted the research approach detailed below.

2.2.1 Literature Resources

To carry out an exhaustive search and cover a wide

range of information in the literature, we identified

candidate primary articles to answer the research

questions. The selection of these articles was

performed by applying search strings in the following

seven digital databases: Scopus, Springer Link, ACM

Digital Library, Science Direct, Wiley, IEEE Xplorer

and Web Of Science.

The search interval was limited between 2010 and

2022. Searches were conducted separately based on

title, abstract, and keywords only for Scopus. For the

other six databases, the search was extended to the

entire article. Note that the searches must be adjusted

according to the needs of each database.

2.2.2 Search Terms

a- Pico

According to the PICO (Population, Intervention,

Comparison and Outcome) criteria suggested by

(Kitchenham & Charters, 2007) and based on the

main terms extracted from RQs and the semantically

similar terms, the search string was defined as

follows:

("Multimodal Biometric" OR "Multi-Modal

Biometric" AND ("Recognition" OR

"Authentication" OR "Identification" OR

"Verification") AND ("System" OR "Scheme" OR

"Approach" OR "Method" OR "Algorithm" OR

"Technic" OR "Model") AND ("Physiological traits"

OR "Finger Vein" OR "Palm Vein" OR "Finger Print"

Or "Face" OR "Lips" OR "Iris" OR "Retina").

Population: It refers to specific software engineering

role, category of software engineer, an application

area or an industry group (Kitchenham & Charters,

2007). In our context, the population are Multimodal

Biometric Systems based on physiological traits;

Intervention

: It refers to a software methodology,

tool, technology, or procedure (Kitchenham &

Charters, 2007). In this study, we focus on recognition

of morphological biometric images;

Comparison

: It refers to the software engineering

methodology /tool/ technology/ procedure with which

the intervention is being compared (Kitchenham &

Charters, 2007). In this study, it intends all terms

related to the biometric recognition process;

Outcomes

: No measurable results are taken into

account, since we do not evaluate the results obtained

in the different articles studied.

b- Adapted Search String:

In order to refine the search in certain electronic

databases, it was necessary to adapt the search string

and some of the associated filters. It should be noted

that the search in these databases was carried out on

09/02/2022.

2.2.3 Search Process

In order to ensure that no candidate article was

eliminated, a research process was followed as

outlined below:

1st step: The first author conducts search in the seven

predefined databases to find an initial batch of

candidate papers, then carefully reviews the title,

abstract, and keywords using the inclusion and

exclusion criteria. In case or information contained in

the metadata is not sufficient to accept or reject the

paper, the full-text must be explored.

2nd step: On a shared spreadsheet, the same author

stores extracted data from selected studies and for

each article he indicates his decision to accept or reject

the paper. In case of doubt, he must imperatively

ICSOFT 2023 - 18th International Conference on Software Technologies

558

mention that he is uncertain in order to open the

discussion with the other authors until they agree.

3rd step: To reduce threats, the other two authors

evaluate independently the validated selection that has

been stored in the previous shared spreadsheet and

review the rejected articles to ensure that no potential

article is excluded. In case of disagreement, a meeting

between the three authors is scheduled to make a final

decision.

2.3 Study Selection Procedure

In this section, we determine the filtering process

applied to the potentially relevant articles. For this, we

describe the following list which specify inclusion

and exclusion criteria used in this study combined by

the OR boolean operator. The number of included and

excluded articles is shown in figure 2 at each stage.

 Inclusion criteria

IC1: The publication format is a peer-reviewed

academic journal or conference paper.

IC2: The study was published online during the period

2010 until 2022.

IC3: The paper develops new and/or use existing

image processing techniques for Multimodal

Biometric Recognition Systems.

 Exclusion criteria

EC1: The paper is a duplicate found in another source

of publication.

EC2: The paper is not in English language.

EC3: The paper cites Multimodal Biometric

Recognition Systems just as an example.

EC4: The paper is available as abstract and/or

PowerPoint presentation.

EC5: The study focuses on Continuous Multimodal

Biometric Authentication Systems.

EC6: The paper is a survey, systematic mapping

study, systematic literature review, a web page or a

poster.

2.4 Data Extraction and Synthesis

In order to extract information from the primary

selected studies, the model described in Table 1 was

developed. The extraction was performed by the first

author and reviewed independently by the second and

third authors. The latter read the full text of all

selected articles and collected the data necessary to

answer the research questions addressed in this article.

2.5 Threats to Validity

The threats to the validity of this work mainly relate

to the exclusion of relevant articles, publication bias

and data extraction bias. One of the main problems we

faced was to find all articles that addressed the

research questions in order to minimize the threat of

exclusion of relevant articles. To achieve this

objective, we searched the seven electronic databases

listed in section 2.2.1, using a search string adapted to

their search engines while respecting the main search

terms and their semantic similarities. However, it is

probable that some relevant studies were not returned

by the search terms we used. To reduce this threat, we

used "Backward Snowball Sampling"; this means that

a manual check of the reference list of each of the

relevant studies to identify those that were missed

during the automated search. To further reduce the

risk of incorrectly excluding relevant articles, two

researchers conducted the process of selecting

relevant studies separately, using inclusion and

exclusion criteria based on title, abstract and

keywords. If in doubt, the full article has been read. In

case of disagreement, a meeting between the three

authors is scheduled to make a final decision. Besides

searching and selecting all relevant studies, data

extraction was the most crucial task of this study. To

properly extract data from these studies, two

researchers read each article independently. The data

extracted for each article was compared and any

disagreements were discussed by the researchers.

3 MAPPING RESULTS

This section presents and discusses the results related

to the systematic mapping questions presented

previously. Note that Figures 3, 4 and 5 were extracted

using Biblioshiny (Aria & Cuccurullo, 2017).

3.1 Overview of the Selected Studies

Figure 2 illustrates the number of articles returned at

each stage of the selection process. We notice that the

search of the seven electronic databases has generated

1289 candidate articles. The application of inclusion

and exclusion criteria served to identify those that

were relevant, because many of the articles would not

be valuable in answering the research questions. This

procedure generated 247 articles. Examination of the

reference lists of the selected articles did not reveal

any other relevant article.

Table 2 indicates the number of studies published by

channel for the period from 2010 to February 2022.

Multimodal Biometric Recognition Systems based on Physiological Traits: A Systematic Mapping Study

559

They were published in different sources, mostly

journals or conferences. 50.79% (123 papers) were

presented in conferences, 45.63% of the studies (115

papers) in journals, 1.98% (5 papers) in symposiums

and 1.19% (3 papers) in workshops; the last paper was

a book chapter.

According to figure 3, Expert Systems with

Applications, Pattern Recognition and the

International Journal of Biometrics top the list of

publication sources for the journal category with a rate

that reaches 5.21%, 4.35% and 3.48% respectively .

As for the conferences, just International Conference

On Trends in Electronics and Informatics (ICEI 2017)

which has a rate of 2.43% (3 conference papers out of

123) and for the others each of them has published two

papers (IET Biometrics, Neurocomputing, Procedia

Computer Science….).

3.2 MQ1: How Publications Are

Distributed across Countries?

The main objective of this question is to identify

universities and research laboratories that are

interested in the subject, for this, we considered the

country of the first author's university. Indeed, as

confirmed in figure 4, India tops the ranking with 32%

of publications and about 246% more than its

successor: China (13% of publications). Next comes

Malaysia with a rate of 2%. The remaining 53% of

publications are distributed among other countries

(Canada, Algeria, USA, Cyprus, Italy, etc.) with a rate

not exceeding 1.5%.

3.3 MQ2: What Is the Quality of

Relevant Articles According to

Conference and/or Journal Rank?

We chose to assess the relevance of each selected

paper according to Scimago Journal & Country Rank

for the scientific journals providing SJR and H Index

scores, and CORE Conference Ranking for the

conferences. In order to standardize the ranking

between journals and conferences, we have calculated

a new score according to the following nomenclature:

 For journals: (+5) if the journal ranking is Q1, (+4)

if the journal ranking is Q2, (+3) if the journal ranking

is Q3, (+2) if the journal ranking is Q4, and (+1) for

others.

 For conferences: (+5) if the conference is CORE A,

(+4) if the conference is CORE B, (+3) if the

conference is CORE C, (+2) if the conference is

CORE D, (+1) if the conference is not ERA ranked

but according to Qualis, and (+0) for others.

We found that 49% of the articles are published in

ranked journals and conferences and the remaining

51% in unranked sources.

3.4 MQ3: According to MQ1 and MQ2,

Is There a Correlation between the

Quality of the Selected Articles and

the Scientific Production?

From the results obtained in section 3-2, we could see

that the scientific production rate in the Asian

continent was very high (47% of all selected articles)

compared to other parts of the world. This led us to

seek more information about the quality of these

works. For this, we calculated an average score for

each country based on the scores obtained for the

research question MQ2, the results are presented in

Table 3. It seems that the number of articles published

per country is not the only criterion to determine the

interest of a laboratory or a university for a given

subject. If we take the example of India and Turkey,

we can deduce that during the period (2010 - February

2022), Turkey published only 4 papers that met the

selection criteria for this work, with an average score

of 4.25; while India has been able to publish an

interesting number of articles in this field, but the

average score is only 1.5.

This observation led us to further examine the

reasons for the interesting increase in the number of

articles published in India. The following was

observed; As we all know; India is the second most

populated country in the world (with 1.41 billion

inhabitants in 2022). In 2010, The Indian government

established the largest biometric database in the

world, called "Aadhaar", managed by the Unique

Identification Authority of India. The system includes

a 12-digit national identification number associated

with each person in addition to biometric data,

including iris image, facial image and fingerprints.

The project also incorporates more standard data, such

as name, gender, date and place of birth (Chander &

Kush, 2010).

Therefore, to succeed in this huge project, unique

in the world, the commitment of the scientific

community was indispensable. This could explicitly

explain the peak of scientific production in India in

this field; the results deduced above place India at the

top of the list for this specific period.

3.5 MQ4: What Is the Annual Trend of

Publications?

Answering this question will give researchers the

opportunity to properly position themselves with

ICSOFT 2023 - 18th International Conference on Software Technologies

560

respect to this disciplinary field and to have an overall

idea of the evolution of annual scientific production.

We see in figure 5 that the number of publications

peaked in 2016 and 2019. On the contrary, in 2017, a

decrease was observed.

3.6 MQ5: Where and When Were the

Selected Studies Published?

As noted in section 3-1, 247 studies were selected out

of 1289 candidate papers, of which 131 were

conference papers, 115 were journal articles, and only

one book chapter. The distribution of publications

across libraries and their types are provided in figure

6. IEEEXPLORE and Scopus dominate the ranking

with 38% and 30% respectively; representing 68% of

the total selected articles. The remaining 32% are

distributed as follows: 13% for SpringerLink, 4% for

WOS and ACM; the last 1% is attributed to Wiley.

3.7 MQ6: What Are the Most Used

Biometric Traits Combinations

Studied in the Selected Papers?

After analyzing the results obtained, we decided to

retain only the first six findings for which the number

was significant compared to the others. In fact, it was

observed, as shown in Figure 7, that the Face-Iris and

the Fingerprint-Iris combination are the most

frequent. Followed by the Face-Fingerprint

combination, then an equal ratio for the Face-

Palmprint and the Fingerprint-Fingervein

combination; in the last comes the Ear-Face

combination. We can therefore see that the iris is in

great demand; which can be explained by the fact that

the iris is one of the most accurate biometric features,

with very low false match rates and high processing

speeds in large-scale datasets (Bowyer,

Hollingsworth, & Flynn, 2008); it is an organ of the

eye, situated directly in front of the cornea and in

behind the lens (Ross, 2010). These observations are

reinforced by the complex texture of its stroma that

differs from an individual to another, the perceived

permanence of its discriminant characteristics, its

high universality and its restricted genetic penetration

(Nguyen, Fookes, Jillela, Sridharan, & Ross, 2017).

3.8 MQ7: What Are the Main Image

Processing Techniques Identified in

the Selected Papers?

The image processing techniques used in the data

extraction phase and mentioned in the selected papers

are very varied as mentioned in Figure 8; note that

more than half (52%) opted for the three main

methods, namely the PCA Algorithm (24%), the LBP

(14%) and the Deep CNN (14%); followed by the

Gabor filter (11%). The other methods were less used:

PSO and DCT Algorithm (8%), Minutiae Algorithm

(7%), DWT (6%) and LDA (4%).

3.9 Tables

Table 1: Data extraction form.

Data Ite

Value RQ

DOI

Article Title

Extractor/

Checke

Source Book/conference/

ournal MQ3

Publication Yea

MQ3

Country MQ1

Publication Type

MQ2 Conf and/or journal ran

MQ4

MQ5

MQ6

MQ7

Table 2: Selected Paper’s Venue.

Publication

source

Publication

Type

Number of

Papers

% of Selected

Papers

Conference Conf. paper 123 50.79%

Journal Journal paper 115 45.63%

Symposium Conf. paper 5 1.98%

Workshop Conf. paper 3 1.19%

Table 3: Average Rating of Publications by Country.

Country Study Conf

Journa

Average

Score

India 132 76 56 1,5

China 33 11 22 2,67

Algeria 10 7 3 1,3

Malaysia 8 3 5 2,75

Canada 7 7 - 0,71

USA 6 4 2 1

Morocco 5 3 2 1,8

Korea 5 2 3 1,2

Turkey 4 0 4 4,25

UK 4 2 2 2,75

Italy 4 1 3 2

Egypt 4 3 1 0

Iran 3 1 2 3,33

Tunisia 3 1 2 2,75

KSA 3 0 3 2

Pakistan 2 0 2 4

Portugal 2 2 0 2,5

Multimodal Biometric Recognition Systems based on Physiological Traits: A Systematic Mapping Study

561

Table 3: Average Rating of Publications by Country.(cont.)

Country Study Conf

Journa

Average

Score

Australia 1 0 1 3

Cyprus 1 0 1 3

Sth Africa 1 1 0 1

Hong Kong 1 1 0 0

Irak 1 1 0 0

3.10

Figures

Figure 1: Mapping process.

Figure 2: Study selection process.

Figure 3: Most relevant sources.

Figure 4: Corresponding Author’s Country.

Figure 5: Annual Scientific Production.

Figure 6: Libraries in which selected papers were published.

Figure 7: The most common biometric trait combinations.

ICSOFT 2023 - 18th International Conference on Software Technologies

562

Figure 8: Distribution of the most used Image Processing

techniques.

4 CONCLUSION AND FUTURE

WORK

The purpose of this SMS was to analyze and

synthesize articles dealing with the design of

multimodal biometric recognition systems based on

physiological traits. 247 relevant articles published

between 2010 and February 2022 were selected and

analyzed by year, source and country of publication,

combinations of biometric traits and image processing

techniques employed.

The current work aims to conduct a literature

review in order to compare the performances recorded

by the different image processing techniques and to

discuss in depth the results obtained from these works

while considering the choice of the used biometrics.

REFERENCES

Abo-Zahhad, M., Ahmed, S. M., & Abbas, S. N. (2014).

Biometric authentication based on PCG and ECG

signals: present status and future directions. Signal,

Image and Video Processing, 8, 739-751.

Aria, M. & Cuccurullo, C. (2017) bibliometrix: An R-tool

for comprehensive science mapping analysis, Journal of

Informetrics, 11(4), pp 959-975, Elsevier.

Bharadwaj, S., Vatsa, M., & Singh, R. (2014). Biometric

quality: a review of fingerprint, iris, and face. EURASIP

journal on Image and Video Processing, 2014(1), 1-28.

Bowyer, K. W., Hollingsworth, K., & Flynn, P. J. (2008).

Image understanding for iris biometrics: A survey.

Computer vision and image understanding, 110(2),

281-307.

Chander, S., & Kush, A. (2010). Unique identification

number and e-governance security. International

Journal of Computing and Business Research, 1(1).

Chauhan, S., Arora, A. S., & Kaul, A. (2010). A survey of

emerging biometric modalities. Procedia Computer

Science, 2, 213-218.

Deriche, M. (2008, November). Trends and challenges in

mono and multi biometrics. In 2008 First Workshops on

Image Processing Theory, Tools and Applications (pp.

1-9). IEEE.

Jain, A., Nandakumar, K., & Ross, A. (2005). Score

normalization in multimodal biometric systems. Pattern

recognition, 38(12), 2270-2285.

Kitchenham, B., & Charters, S. (2007). Guidelines for

performing systematic literature reviews in software

engineering.

Nguyen, K., Fookes, C., Jillela, R., Sridharan, S., & Ross,

A. (2017). Long range iris recognition: A survey.

Pattern Recognition, 72, 123-143.

Petersen, K., Feldt, R., Mujtaba, S., & Mattsson, M. (2008,

June). Systematic mapping studies in software

engineering. In 12th International Conference on

Evaluation and Assessment in Software Engineering

(EASE) 12 (pp. 1-10).

Petersen, K., Vakkalanka, S., & Kuzniarz, L. (2015).

Guidelines for conducting systematic mapping studies

in software engineering: An update. Information and

software technology, 64, 1-18.

Ross, A. (2010). Iris recognition: The path forward.

Computer, 43(2), 30-35.

Ross, A., & Jain, A. K. (2004, September). Multimodal

biometrics: An overview. In 2004 12th European signal

processing conference (pp. 1221-1224). IEEE.

Rui, Z., & Yan, Z. (2018). A survey on biometric

authentication: Toward secure and privacy-preserving

identification. IEEE access, 7, 5994-6009.

Shaheed, K., Liu, H., Yang, G., Qureshi, I., Gou, J., & Yin,

Y. (2018). A systematic review of finger vein

recognition techniques. Information, 9(9), 213.

Shaheed, K., Mao, A., Qureshi, I., Kumar, M., Abbas, Q.,

Ullah, I., & Zhang, X. (2021). A systematic review on

physiological-based biometric recognition systems:

Current and future trends.

Archives of Computational

Methods in Engineering, 28(7), 4917-4960.

Unar, J. A., Seng, W. C., & Abbasi, A. (2014). A review of

biometric technology along with trends and prospects.

Pattern recognition, 47(8), 2673-2688.

Multimodal Biometric Recognition Systems based on Physiological Traits: A Systematic Mapping Study

563