Fuzzy Logic Based Edge Detection Methods: A Systematic Literature

Review

Miqu

eias Amorim Santos Silva

1 a

, Gracaliz Dimuro

1 b

, Eduardo Borges

1 c

Giancarlo Lucca

1 d

and Cedric Marco-Detchart

2 e

Center of Computational Sciences, Federal University of Rio Grande, It

alia avenue, Rio Grande - RS, Brazil

Universitat Polit

ecnica de Val

encia, Valencia, Comunitat Valenciana, Spain

Keywords:

Edge Detection, Fuzzy Methods, Systematic Literature Review.

Abstract:

Edge detection, or the detection of the maximum limit between two regions with different properties, is one of

the classic problems in the area of computer vision. The uncertainty associated with the nature of this detec-

tion, such as the characteristic fuzzy transition zone resulting from the image discretization processes, or even

noise and illumination variations, justiﬁes an approach based on fuzzy logic theory. In order to understand the

state of the art in edge detection techniques using fuzzy logic-based methods, this work proposes a systematic

review considering two bibliographic sources of scientiﬁc literature, Scopus and Web Of Science. In total, 34

works were selected through a systematic literature review, and their methods were summarized and reported

in this research. From this analysis, it could be concluded that, in recent years, fuzzy logic has been em-

ployed in hybrid methods in order to improve the performance of existing techniques or reduce computational

complexity. Studies with interval fuzzy logic of higher order have been employed for its greater ﬂexibility in

dealing with the uncertainty associated with the edge detection task.

1 INTRODUCTION

One of the most common approaches for detecting

discontinuities in images is edge detection (Suresh

and Srinivasa Rao, 2019). An edge is deﬁned as the

maximum boundary between two regions with differ-

ent properties (Martin, 2002), i.e. the border between

two objects or object faces in an image.

Edge detection represents an important task in

several steps of computer vision and image process-

ing such as (Wei et al., 2017); object detection (Yang

et al., 2002); pattern recognition (Mohan et al., 2021)

and others. Either because of the discretization intrin-

sic to the digital capture process, or because of some

subsequent quantization process, the edge of the ob-

jects, or the faces of the objects, show a small smooth-

ing around the actual boundary of the regions.

This uncertainty, among other characteristics of

the images, makes it difﬁcult to accurately determine

https://orcid.org/0000-0003-4496-500X

https://orcid.org/0000-0001-6986-9888

https://orcid.org/0000-0003-1595-7676

https://orcid.org/0000-0002-3776-0260

https://orcid.org/0000-0002-4310-9060

the edge of objects, so different methods have been

proposed throughout history, from methods based on

partial derivatives such as Sobel (Sobel et al., 1968),

Log (Marr and Hildreth, 1980) and Canny (Canny,

1986) in the 1970s and 1980s, up to methods based

on convolutional neural networks in recent years (Jing

et al., 2022).

Traditional methods have a number of problems,

such as high sensitivity to noise, high complexity, and

high consumption of time and processing. In gen-

eral, these methods show very discontinuous results

and false positives.

On the other hand, methods based on fuzzy logic

(Zadeh, 1965) were developed in order to represent

the imprecision and uncertainty of the information

(Tripathi et al., 2021). In recent years, there has been

increasing research on applications of fuzzy logic in

several areas such as pattern recognition, neural net-

works, expert systems, artiﬁcial intelligence, control

theory, automata, decision-making, medical diagno-

sis, and robotics. (Muthalagu et al., 2020; Orhei et al.,

2020; Qiu et al., 2021).

In computer vision, in addition to applications in

edge detection (Bustince et al., 2009; Lopez-Molina

et al., 2010; Marco-Detchart et al., 2021b), fuzzy ap-

Silva, M., Dimuro, G., Borges, E., Lucca, G. and Marco-Detchart, C.

Fuzzy Logic Based Edge Detection Methods: A Systematic Literature Review.

DOI: 10.5220/0011853700003467

In Proceedings of the 25th International Conference on Enterprise Information Systems (ICEIS 2023) - Volume 1, pages 385-394

ISBN: 978-989-758-648-4; ISSN: 2184-4992

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

385

proaches have been frequent in noise reduction (Yuk-

sel and Basturk, 2012), feature extraction (Shimada

et al., 2005), classiﬁcation and clustering (Koschan

and Abidi, 2005).

Considering the importance of research in edge

detection through the fuzzy theory approach, and

since, to the best of our knowledge, there are no re-

view papers devoted speciﬁcally to fuzzy methods,

a speciﬁc investigation/review of these methods is

needed

In this sense, the main research question which is

intended to be answered is: RQF - what is the state

of the art of fuzzy edge detection methods? In order

to relate them to the rest of the literature in the area

of edge detection, the following should also be an-

swered: RQG - what is the state of the art of edge

detection methods?

For each research question, a bibliometric analy-

sis was sought in order to provide a set of information

that would appropriately position new research activ-

ities, this being another contribution of this work, be-

sides the summarization and description of the found

methods.

This paper is organized as follows: in Section 2

(Methodology), the search terms, inclusion and exclu-

sion criteria, as well as the number of papers found

and admitted for review will be discussed. Next, in

Section 3, the results and discussion are presented for

each research question mentioned above, summariz-

ing the reviewed methods with their respective biblio-

metric analyses. Finally, in Section 4, the conclusions

are exposed.

2 METHODOLOGY

This section ﬁrst introduces the concept of a system-

atic literature review and then presents the methodol-

ogy used to answer the research problems.

2.1 Systematic Literature Review

According to (Kitchenham and Charters, 2007), a sys-

tematic literature review is a form of analysis that

aims to identify, evaluate, and interpret all avail-

able relevant research on a speciﬁc research problem,

topic, or phenomenon of interest. Among the various

reasons for engaging in a systematic literature review,

the most common are:

i. Summarize the existing evidence regarding treat-

ment or technology;

ii. Identify gaps in current research with the goal of

suggesting areas for future research;

iii. Provide a set of information that appropriately po-

sitions new research activities and;

iv. Reduce, or try to eliminate research bias.

In this context, this research seeks to summarize the

existing technology regarding fuzzy edge detection

methods. For this, key good practice questions for a

quality review, found in (Tacconelli, 2010), were con-

sidered.

2.2 Deﬁnition of Criteria, Search in the

Indexing Bases and Obtaining

Primary Research

In order to answer the research problems, searches

were conducted in two major bibliographic databases,

Scopus (SC)

and Web of Science (WS)

. In total,

34 papers were selected by merging the results of

the searches in the two databases (21 from the SC

database and 16 from the WS database). This union,

by deﬁnition, considers the exclusion of duplicates.

The summary of all the included information and

documents that went through each exclusion step,

which will be discussed below, is shown in Table 1,

where we show for each search question, the con-

sidered inclusion terms, where ABS() refers to the

terms found in the abstract and TITLE() refers to the

terms found in the title. In the number of occurrences,

we have the total number of papers found by each

database, where inc is the number of papers found in

that search, and for each exclusion step (exc) the total

number of elected papers.

One can see in Table (1) that four exclusion steps

were considered:

i. Step one:

a) Exclusion of papers published before the

year 2017.

ii. Step two:

a) Elimination of those that were not written

in the English language;

b) Deletion those that have not been pub-

lished in the computer science or related ﬁeld and;

c) Elimination papers that had a title that

demonstrated the lack of relevance of that study

to this review work.

iii. Step three:

a) Deletion of the works ﬂagged as portrayed

by the journal and;

b) Analysis of the abstract, eliminating arti-

cles that were outside the scope of this review.

www.scopus.com

www.webofscience.com

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

386

iv. Step four:

a) Analysis of the methodology and exclu-

sion of those with no validation of the proposed

segmentation method.

2.3 Bibliometric Analysis

Bibliometric analysis is deﬁned as the quantitative

disclosure of the characteristics of a set of works, with

the objective of managing knowledge and scientiﬁc

information on a speciﬁc theme or subject (Aria and

Cuccurullo, 2017). One can cite as observable param-

eters that are commonly analyzed from the articles

and papers selected: their references, authors, num-

ber of citations, and most relevant journals.

To perform the bibliometric analysis of the papers

in this research, we decided to use the bibliometrix

tool through the Rstudio software (Aria and Cuccu-

rullo, 2017). Each set of papers, which were selected

in order to answer one of the problems of this re-

search, was submitted to bibliometric analysis, evalu-

ating the keywords, authors, afﬁliation, abstract, cita-

tions, and country of origin, reporting only the infor-

mation relevant to the research problem.

3 RESULTS AND DISCUSSION

This section presents the results of each bibliomet-

ric analysis of each group of papers, the summary of

the methods found in each set of papers, and a short

discussion aiming to synthesize the obtained informa-

tion.

3.1 State of the Art in Edge Detection

(RQG)

In order to identify the state of the art of edge detec-

tion methods in the literature, we have chosen the pa-

pers that are summarized in Table 2. The biliometry

performed aimed to understand the frequent terms, as

well as those that have gained relevance in the last 6

years.

First, the absolute numbers of citations that each

work received were extracted from the metadata, to

determine those with greater relevance to the research

topic. In addition, the most cited papers are those that,

potentially, best summarize the state of the art until

that year of publication, and have a higher ratio be-

tween the number of citations and the time elapsed

since publication.

In this case, the articles (Magnier et al., 2018;

Kaur and Kaur, 2017; Yogesh et al., 2018) appear as

the most cited, in the ﬁrst three positions. Evaluating

the number of citations/year, from publication to the

present moment, we observe that the works (Suresh

and Srinivasa Rao, 2019; Magnier et al., 2018) have

the highest scores. In the case of (Suresh and Srini-

vasa Rao, 2019) it is the most recent work with the

highest citation/year ratio.

Through the word cloud of the keywords of all the

papers, if we disregard terms such as: “edge detec-

tion”, “segmentation”, “image processing” and other

terms that are synonyms or homonyms to them since

they are expected words considering the search terms,

we ﬁnd related and recurrent terms that may indicate

the methods or applications that have been developed

in the studied period. Figure 1 summarizes the most

frequent terms found, among them, we can observe

terms related to fuzzy logic, machine learning, ge-

netic algorithms, and other classical segmentation and

edge detection techniques.

Figure 1: Word clouds of key terms of RQG edge detection

methods.

The occurrence of terms related to classical seg-

mentation and edge detection techniques such as

Canny, Sobel, and Watersheed in the keywords may

indicate a movement towards improving the perfor-

mance of these techniques that have already been ex-

haustively studied. Further evidence of this could be

found through another analysis, where the terms were

considered by year. In this case, words like “edge”,

“detection” and “segmentation” were the most fre-

quent in the abstracts of the papers in the year 2020,

where performance-related terms also started, which

had a higher frequency in the year 2022. From these

papers, it is interesting to extract especially the meth-

ods reviewed by the authors regarding the state of the

art in the area of edge detection. Therefore, in the fol-

lowing, we will present a summary of the techniques,

grouped by approach.

Gradient-Based Methods: A digital image is a dis-

crete representation of the variation of light in the

real world, so each numerical value carried by the

pixel represents the intensity of light or color at that

Fuzzy Logic Based Edge Detection Methods: A Systematic Literature Review

387

Table 1: Terms of inclusion and exclusion and number of selected papers.

Research question Terms of Inclusion

Number of occurrences

SCOPUS WEB OF SCIENCE

RQG

(ABS( edge PRE/ detection )

ABS(image PRE/ segmentation ))

AND

TITLE(survey or review)

inc: 768

1ª exc: 518 (i)

2ª exc: 244 (ii)

3ª exc: 15 (iii)

4ª exc: 15 (iv)

inc: 1,475

1ª exc: 1,038(i)

2ª exc: 248(ii)

3ª exc: 3(iii):

4ª exc: 3 (iv)

RQF

ABS ( fuzzy

AND

( edge PRE/ detection ) )

AND

TITLE ( fuzzy OR edge OR segmentation )

inc: 870

1ª exc: 247 (i)

2ª exc: 7 (ii)

3ª exc: 6 (iii)

4ª exc: 6(iv)

inc: 717

1ª exc: 224 (i)

2ª exc: 21(ii)

3ª exc: 13(iii)

4ª exc: 13(iv)

Table 2: Papers in edge detection (RQG).

Reference Paper title

(Kaur and Kaur, 2017) An Edge detection technique with image segmentation using Ant Colony Optimiza-

tion: A review

(Castillo et al., 2017) Review of Recent Type-2 Fuzzy Image Processing Applications

(Yogesh et al., 2018) A comparative review of various segmentation methods and its application

(Magnier et al., 2018) A review of supervised edge detection evaluation methods and an objective compari-

son of ﬁltering gradient computations using hysteresis thresholds

(Magnier, 2018) Edge detection: a review of dissimilarity evaluations and a proposed normalized mea-

sure

(Agrawal and Bhogal, 2019) A review—Edge detection techniques in dental images

(Aggarwal et al., 2019) Review of Segmentation Techniques on Multi-Dimensional Images

(Zhu and Li, 2019) Survey on the image segmentation algorithms

(Suresh and Srinivasa Rao,

2019)

Various image segmentation algorithms: A survey

(Ghosh et al., 2020) Different EDGE Detection Techniques: A Review

(Budzyn and Rzepka, 2020) Review of edge detection algorithms for application in miniature dimension measure-

ment modules

(Chakrapani et al., 2021) A Survey of Sobel Edge Detection VLSI Architectures

(Isa et al., 2021) Review of Edge-based Image Segmentation on Electrical Tree Classiﬁcation in Cross-

linked Polyethylene (XLPE) Insulation

(Mubashar et al., 2022) Have We Solved Edge Detection? A Review of State-of-the-art Datasets and DNN

based Techniques

(Yadav and Pandey, 2022) Image Segmentation Techniques: A Survey

(Jing et al., 2022) Recent advances on image edge detection: A comprehensive review

point. The contours of objects in the image can be

interpreted as a transition zone between these inten-

sities so that more intense transitions have a much

greater chance of being an edge than smoother transi-

tions. That is, given a direction in the image, the rate

of change or discrete difference between pixels high-

lights the likely edges of the objects (Marco-Detchart

et al., 2021a).

It was through this reasoning that the ﬁrst edge de-

tectors emerged (Agrawal and Bhogal, 2019). There

are two approaches to gradient-based detection: ﬁrst-

order derivative-based, and second-order derivative-

based.

The best-known gradient-based detection methods

are ﬁrst-order ﬁxed operations; ﬁrst-order oriented

operations that use the maximum energy of the ori-

entation; or two-direction operations.

To facilitate understanding regarding the develop-

ment of the techniques throughout history, we have

summarized in a timeline the publication dates of the

articles describing the ﬁrst version of each method de-

scribed in this section, which can be seen in Figure.

2. The following are some of the classical gradient-

based methods and their main differences (Aggarwal

et al., 2019).

The Canny Detector (Canny, 1986), to this day

one of the most widely used, was proposed consid-

ering a Gaussian smoothing followed by a gradient

operation and ﬁnally thresholding. In the studies in

(Jing et al., 2022), we found other methods based

on ﬁrst-order derivatives, inspired by Canny, such

as Inﬁnite size Symmetrical Exponential Filters (D-

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

388

ISEF), a color-boundary and ﬁrst-order derivative of

anisotropic Gaussian (FDAG).

Sobel’s detector (Sobel et al., 1968) calculates the

gradient value at each pixel position in the image us-

ing a ﬁxed operator. In turn, Prewitt (Prewitt et al.,

1970) has a similar technique to Sobel, but unlike So-

bel it has no adjustment coefﬁcients, varies by a con-

stant value, and calculates the magnitude of the gradi-

ent with the image orientations.

The method proposed by Roberts (Roberts, 1980)

ﬁrst appears in the literature in 1963, as the product of

his doctoral thesis. This method calculates the deriva-

tive by taking the root of the difference between diag-

onally adjacent pixels. It focuses on invariant proper-

ties that edges exhibit.

The LoG (Marr and Hildreth, 1980), or Gaussian

Laplacian, is based on a Gaussian ﬁltering followed

by a Laplacian operation. It is based on a second-

order operator, which seeks to identify the maximum

and minimum points of the variation of intensities so

that by ﬁnding these points that cross to zero, the edge

candidates in the image are found. The Gaussian ﬁlter

step is important because it is a second-order method

and therefore extremely sensitive to noise.

Region Segmentation-Based Methods: Another set

of methods for edge detection is based on region seg-

mentation. In these methods, regions are segmented,

such as clustering and automated thresholding meth-

ods, and edges are detected as the boundary of these

regions. According to (Mubashar et al., 2022), the re-

gions formed by the textures in high-complexity im-

ages can be used as a facilitator in the process of de-

tecting the edges between them.

The region-based approach outperforms direct de-

tection methods such as gradient methods. Others,

such as frequency domain ﬁltering and statistics can

be classiﬁed as either region segmentation or direct

edge detection methods, depending on how the detec-

tion modeling is done. These works can be found in

(Jing et al., 2022; Mubashar et al., 2022).

Through the emergence of texture descriptors and

other local information such as brightness gradient,

texture gradient, and color gradient, the probabilistic

contour (Pb) method emerged (Arbelaez et al., 2010).

Methods Based on Machine Learning and Neural

Networks: Combining the Pb method with a logis-

tic regression it was possible to develop a model for

edge detection in the image (Jing et al., 2022). The

method has been extended over the years, for example

by bringing in the multi-scale probabilistic contouring

method and a multi-scale spectral clustering.

Currently, new machine learning-based tech-

niques have emerged, in particular those based on

Convolutional Neural Networks (CNN’s). Other

methods based on machine learning are presented in

(Jing et al., 2022), this is the case of Holistically-

nested edge detection better known as HED that was

proposed to improve the performance of the convo-

lutional neural network, inspired other methods as,

Convolutional Encoder-Decoder Network (CEDN),

Richer Convolutional Features (RCF), Learning to

Predict Crisp Boundaries (LPCB).

In 2019, the bi-directional cascade network

(BDCN) method emerges, which proposes detection

at different scales. More recently, techniques such as:

Fined, Edge Detection TransformER (EDTER), and

Pixel Difference Networks (PiDiNet) have been pro-

posed, to handle edge detection without the need for

such a large database for model training.

Fuzzy Logic Based Methods: Since Russo (Russo,

1998) ﬁrst presented a fuzzy inference system model-

ing to efﬁciently extract edges in high noise images in

1998, the application of fuzzy theory in edge detec-

tion has increased, justiﬁed by the fuzzy nature of ob-

ject edges in a digital image, which makes fuzzy the-

ory suitable for solving such problems (Russo, 1998).

As reported by (Jing et al., 2022), detectors based

on joining techniques like divergence and fuzzy en-

tropy minimization (FED) and detectors based on

morphological gradient and type-2 fuzzy logic have

been proposed (Type-2 Color). Hybrid techniques

and fuzzy versions of neural networks have been cre-

ated, improving the performance of other approaches

and in some cases decreasing the computational com-

plexity.

In the studies of (Ghosh et al., 2020) we can

identify hybrid methods that use neural networks

and fuzzy logic type-1, type-2, and type-3, which

are extensions of fuzzy logic with more degrees of

uncertainty associated, they are called Hybrid ap-

proach neuro-fuzzy-1, neuro-fuzzy-2 and neuro-fuzzy-

3 (Neuro-fuzzy’s). As well as the use of fuzzy

logic to improve classical methods such as Canny

(C&I-TYPE2), and a hybrid method that uses Sobel,

fuzzy logic type-1, and fuzzy interval system type-2

(T2FLS).

Figure 3 presents the chronology of the found

fuzzy methods, through the two proposed research

questions. Observe the predominance of applications

of fuzzy logic type-1, type-2, and type-3 and the use

of this approach in conjunction with other techniques,

evidencing a clear research trend.

3.2 Fuzzy Logic-Based Edge Detection

Methods (RQF)

Considering the current state of edge detection meth-

ods and the importance of fuzzy logic-based methods,

Fuzzy Logic Based Edge Detection Methods: A Systematic Literature Review

389

1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020

Roberts

sobel and Prewitt

LOG

Canny

D-ISEF

Color-boundary

HED

CEDN

RCF and FDAG

LPCB

Fined

PiDiNet

EDTER

Figure 2: Chronology of non-fuzzy methods RQG.

1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022

Russo Method

C&I-TYPE2

LFIT-n & FHM & FISM & FCEM

Neuro-fuzzy’s

T2FLS

FED

Type-2 Color

Figure 3: Chronology of fuzzy methods RQF.

evidenced in the previous section, it was possible to

perform a bibliometric analysis of these articles in or-

der to obtain some useful information in understand-

ing the state of the art of fuzzy logic-based methods.

It is important to emphasize that some works fo-

cused on the review of segmentation methods, in gen-

eral, using fuzzy logic, were included in this review.

This is justiﬁed by the fact that the process of edge de-

tection is a method of segmentation of the boundary

between two or more regions, i.e., a particular case of

the segmentation area in general and, therefore, high-

impact works that were not speciﬁc to edge detection

but returned from consultations and went through the

exclusion processes were admitted.

Table 3 summarizes the papers returned from the

search, organized by publication date. In the ﬁrst col-

umn, we show the reference and, next to it, the title of

the work.

Next, we present the relevant results of the biblio-

metric analysis, with the objective of presenting the

most important papers and the common terms among

the papers. This may allow a better understanding of

the research carried out in this area, with the help of

the description of the methods, to understand the gaps

in the literature regarding this theme.

By extracting the metadata of the papers, it was

possible, as described in the previous chapter for gen-

eral edge detection methods, to determine the most

cited papers among the selected ones. The paper with

the highest number of absolute citations was the paper

by (Castillo et al., 2017). In the second and third po-

sitions, in absolute numbers of citations, we have the

works (Kumar et al., 2019; Gonzalez et al., 2017a).

Analyzing the ratio of citations per year, we have

that the three most cited papers in absolute number,

are also those with the highest citations/year ratio

without losing the hierarchical order. Of these works,

two are about edge detection using type-2 interval

fuzzy logic, and the second position is relatively re-

cent.

Figure 4: Word cloud of key terms of fuzzy methods.

Through the word cloud, it is possible to iden-

tify, disregarding obvious terms such as those used

in the search, or synonyms and homonyms, the terms

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

390

that have raised the interest of researchers in the area,

when using a fuzzy treatment in solving the prob-

lem of edge detection and image segmentation. In

Figure 4, one can observe terms such as: “Medi-

cal Imaging”, “Computer Circuits”, “General Type-2

fuzzy sets”, “Classical Methods” and “Sobel Opera-

tor”. This word cloud was obtained through the fre-

quency of the terms in the keywords, which are terms

chosen by the authors to describe their work, using

the software R Studio.

The metrics presented in the results of these works

in comparison with classical methods show that the

approach plays a key role in improving the perfor-

mance of classical and other existing methods.

The found methods can be divided into four

classes:

i. Methods based on interval fuzzy logic of the type-

n (LFIT-n) (Baghbani et al., 2019; Raheja and

Kumar, 2021; Gonzalez et al., 2017a; Kaur and

Kaur, 2017; Castillo et al., 2017; Gonzalez et al.,

2017b);

ii. Hybrid methods (FHM), in combination with

neural networks or improving the performance

of classical edge detection techniques (Balaban-

taray et al., 2017; Kumawat and Panda, 2021;

Moya-Albor et al., 2017; Dhargupta et al., 2019;

Kaur and Kaur, 2017; Dorrani et al., 2020);

iii. Methods that are based on fuzzy inference sys-

tems (FISM) (Dhivya and Prakash, 2019) e;

iv. Methods that are based on fuzzy clustering meth-

ods (FCEM) (Flores-Vidal et al., 2018).

4 CONCLUSION

Edge detection represents an important task in several

steps in computer vision, such as (Wei et al., 2017);

object detection (Yang et al., 2002); pattern recogni-

tion (Mohan et al., 2021); or image retrieval (Pavithra

and Sharmila, 2018).

Recently, research in this area has been gaining

momentum due to its importance in several applica-

tions, such as augmented reality (Orhei et al., 2020);

image colorization (Sun et al., 2019) and medical im-

age processing (Qiu et al., 2021) among others. Be-

cause of the fuzzy nature of the edges of an object

in a digital image, many methods have been devel-

oped over the years, and more recently, with the use

of fuzzy theory, which has been improving and per-

forming applications in various areas.

Considering the advancement of research in edge

detection through fuzzy theory approaches, and the

lack of review papers dedicated to these methods, a

speciﬁc investigation of these techniques is justiﬁed.

In this sense, this work presented a review of

edge detection methods in order to understand the

state of the art of fuzzy edge detection methods. For

this, searches were performed in two major databases,

Scopus and Web Of Science, and deﬁned the terms

of inclusion and exclusion, so that of 3,830 papers

that returned from the searches, only 34 went through

the exclusion steps. The methods were reviewed and

organized by category according to the proposed ap-

proach by the method.

Classical methods can be divided into four classes:

methods based on gradient; methods based on seg-

mentation of regions; methods based on machine

learning and Neural Networks and; methods based on

Fuzzy Logic that in turn, through the speciﬁc search

can be divided into four subclasses: methods based on

interval fuzzy logic type-n; hybrid methods, in com-

bination with neural networks or improving the per-

formance of classical techniques for edge detection;

those based on fuzzy inference systems and; based on

clustering methods.

The fuzzy set theory proposed by Zadeh (Zadeh,

1965) as well as the Roberts method (Roberts, 1980)

of edge detection, has its ﬁrst works in the mid-1970s.

Despite the nature of edge detection ﬁtting clearly in a

fuzzy theory approach, fuzzy applications in this task

only appeared for the ﬁrst time in the late 1990s using

fuzzy inference systems, with the method of Russo

(Russo, 1998).

Even after this initiative, little has been explored

over the years, with the ﬁrst relevant papers around

the year 2012. A bibliometric analysis was also con-

ducted, where the metadata of the documents was an-

alyzed, in order to understand, in conjunction with the

summary of the methods whether there are gaps or

trends in research regarding the studied topic.

Through the analysis, it could be concluded that in

recent years, from the year 2012, fuzzy logic has been

employed in hybrid methods in order to improve the

performance of existing techniques or reduce compu-

tational complexity. Studies with interval fuzzy logic

of higher order have been employed for its greater

ﬂexibility in dealing with the uncertainty associated

with the edge detection task.

As future work, a review dedicated to the perfor-

mance of the methods is proposed, comparing works

that used the same metrics and databases in order to

evaluate the gain that these different techniques have

in relation to other non-fuzzy methods. This quan-

titative evaluation is necessary considering the num-

ber of methods proposed in the last ten years, and the

scarcity of review work in this regard.

Fuzzy Logic Based Edge Detection Methods: A Systematic Literature Review

391

Table 3: Works in edge detection based on fuzzy logic (RQF).

References Paper title

(Gonzalez et al., 2017b) Edge detection methods based on generalized type-2 fuzzy logic systems

(Castillo et al., 2017) Review of Recent Type-2 Fuzzy Image Processing Applications

(Moya-Albor et al., 2017) An Edge Detection Method using a Fuzzy Ensemble Approach

(Gonzalez et al., 2017a) Edge detection method based on general type-2 fuzzy logic applied to color images

(Balabantaray et al., 2017) A Quantitative Performance Analysis of Edge Detectors with Hybrid Edge Detector

(Flores-Vidal et al., 2018) A New Edge Detection Approach Based on Fuzzy Segments Clustering

(Bueno et al., 2018) Two-phase ﬂow bubble detection method applied to natural circulation system using

fuzzy image processing

(Baghbani et al., 2019) A method for image edge detection based on interval-valued fuzzy sets

(Flores-Vidal et al., 2019) A new edge detection method based on global evaluation using fuzzy clustering

(Dhargupta et al., 2019) Fuzzy edge detection based steganography using modiﬁed Gaussian distribution

(Bhogal and Agrawal, 2019) Image Edge Detection Techniques Using Sobel, T1FLS, and IT2FLS

(Kumar et al., 2019) Information hiding with adaptive steganography based on novel fuzzy edge identiﬁ-

cation

(Dhivya and Prakash, 2019) Edge detection of satellite image using fuzzy logic

(Dorrani et al., 2020) Image Edge Detection with Fuzzy Ant Colony Optimization Algorithm

(Kumawat and Panda, 2021) A robust edge detection algorithm based on feature-based image registration (FBIR)

using improved canny with fuzzy logic (ICWFL)

(Raheja and Kumar, 2021) Edge detection based on type-1 fuzzy logic and guided smoothening

(Tripathi et al., 2021) Edge Detection on Medical Images Using Intuitionistic Fuzzy Logic

ACKNOWLEDGEMENTS

This work was supported by CNPq (301618/2019-

4, 305805/2021-5), FAPERGS (19/2551-0001279-9,

19/2551-0001660, Edital FAPERGS/CNPq 07/2022

- Programa de Apoio

a Fixac¸

ao de Jovens Doutores

no Brasil, Conseller

ıa d’Innovaci

o, Universitats,

Ciencia I Societat Digital from Comunitat Valen-

ciana (APOSTD/2021/227) through the European So-

cial Fund (Investing In Your Future), and grant

PID2021-123673OB-C31 funded by MCIN/AEI/

10.13039/501100011033 and by “ERDF A way of

making Europe”.

DECLARATIONS

The authors declare that there is no ﬁnancial or per-

sonal conﬂict directly or indirectly related to this

work.

REFERENCES

Aggarwal, P., Mittal, H., Samanta, P., and Dhruv, B.

(2019). Review of Segmentation Techniques on

Multi-Dimensional Images. In 2018 International

Conference on Power Energy, Environment and Intel-

ligent Control, PEEIC 2018, pages 268–273. Institute

of Electrical and Electronics Engineers Inc.

Agrawal, A. and Bhogal, R. (2019). A review—Edge de-

tection techniques in dental images. Lecture Notes

in Computational Vision and Biomechanics, 30:1359–

1378. Publisher: Springer Netherlands.

Arbelaez, P., Maire, M., Fowlkes, C., and Malik, J. (2010).

Contour detection and hierarchical image segmenta-

tion. IEEE transactions on pattern analysis and ma-

chine intelligence, 33(5):898–916.

Aria, M. and Cuccurullo, C. (2017). bibliometrix : An

R-tool for comprehensive science mapping analysis.

Journal of Informetrics, 11(4):959–975.

Baghbani, R., Jamasb, S., and Khodabakhshi, M. B.

(2019). A method for image edge detection based

on interval-valued fuzzy sets. JOURNAL OF IN-

TELIGENT & FUZZY SYSTEMS, 37(2):2275–2288.

Place: NIEUWE HEMWEG 6B, 1013 BG AMS-

TERDAM, NETHERLANDS Publisher: IOS PRESS

Type: Article.

Balabantaray, B. K., Sahu, O. P., Mishra, N., and Biswal,

B. B. (2017). A Quantitative Performance Analy-

sis of Edge Detectors with Hybrid Edge Detector.

JOURNAL OF COMPUTERS, 12(2):165–173. Place:

PO BOX 40, OULU, 90571, FINLAND Publisher:

ACAD PUBL Type: Article.

Bhogal, R. K. and Agrawal, A. (2019). Image Edge De-

tection Techniques Using Sobel, T1FLS, and IT2FLS.

In SC., S. and A., J., editors, INFORMATION AND

COMMUNICATION TECHNOLOGY FOR INTEL-

LIGENT SYSTEMS, ICTIS 2018, VOL 2, volume

107 of Smart Innovation Systems and Technologies,

pages 303–317, GEWERBESTRASSE 11, CHAM,

CH-6330, SWITZERLAND. SPRINGER INTERNA-

TIONAL PUBLISHING AG. ISSN: 2190-3018 Type:

Proceedings Paper.

Budzyn, G. and Rzepka, J. (2020). Review of edge detec-

tion algorithms for application in miniature dimension

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

392

measurement modules. Journal of Machine Engineer-

ing, 20(4):74–85. Publisher: Editorial Institution of

Wrocaw Board of Scientiﬁc.

Bueno, R., Masotti, P., Justo, J., Andrade, D., Rocha, M.,

Torres, W., and de Mesquita, R. (2018). Two-phase

ﬂow bubble detection method applied to natural circu-

lation system using fuzzy image processing. Nuclear

Engineering and Design, 335:255–264. Publisher: El-

sevier Ltd.

Bustince, H., Barrenechea, E., Pagola, M., and Fernandez,

J. (2009). Interval-valued fuzzy sets constructed from

matrices: Application to edge detection. Fuzzy Sets

and Systems, 160(13):1819–1840.

Canny, J. (1986). A computational approach to edge de-

tection. IEEE Transactions on Pattern Analysis and

Machine Intelligence, PAMI-8(6):679–698.

Castillo, O., Sanchez, M. A., Gonzalez, C. I., and Martinez,

G. E. (2017). Review of Recent Type-2 Fuzzy Im-

age Processing Applications. INFORMATION, 8(3).

Place: ST ALBAN-ANLAGE 66, CH-4052 BASEL,

SWITZERLAND Publisher: MDPI Type: Review.

Chakrapani, S., Y., Rao, V., N., and Kamaraju, M. (2021). A

Survey of Sobel Edge Detection VLSI Architectures.

In A.J., O., S., A., B.H., H., and H.A., A., editors,

Journal of Physics: Conference Series, volume 1804.

IOP Publishing Ltd. ISSN: 17426588 Issue: 1.

Dhargupta, S., Chakraborty, A., Ghosal, S. K., Saha, S.,

and Sarkar, R. (2019). Fuzzy edge detection based

steganography using modiﬁed Gaussian distribu-

tion. MULTIMEDIA TOOLS AND APPLICATIONS,

78(13):17589–17606. Place: VAN GODEWIJCK-

STRAAT 30, 3311 GZ DORDRECHT, NETHER-

LANDS Publisher: SPRINGER Type: Article.

Dhivya, R. and Prakash, R. (2019). Edge detection of

satellite image using fuzzy logic. Cluster Computing,

22:11891–11898. Publisher: Springer.

Dorrani, Z., Farsi, H., and Mohamadzadeh, S. (2020).

Image Edge Detection with Fuzzy Ant Colony Op-

timization Algorithm. INTERNATIONAL JOUR-

NAL OF ENGINEERING, 33(12):2464–2470. Place:

16, AHURAMAZDA ST, ALVAND AVE, TEHRAN,

15169, IRAN Publisher: MATERIALS & ENERGY

RESEARCH CENTER-MERC Type: Article.

Flores-Vidal, P. A., Gomez, D., Olaso, P., and Guada, C.

(2018). A New Edge Detection Approach Based on

Fuzzy Segments Clustering. In Kacprzyk, J., Szmidt,

E., Zadrozny, S., Atanassov, K., and Krawczak,

M., editors, ADVANCES IN FUZZY LOGIC AND

TECHNOLOGY 2017, VOL 2, volume 642 of Ad-

vances in Intelligent Systems and Computing, pages

58–67, GEWERBESTRASSE 11, CHAM, CH-6330,

SWITZERLAND. SPRINGER INTERNATIONAL

PUBLISHING AG. ISSN: 2194-5357 Type: Proceed-

ings Paper.

Flores-Vidal, P. A., Olaso, P., Gomez, D., and Guada, C.

(2019). A new edge detection method based on global

evaluation using fuzzy clustering. SOFT COMPUT-

ING, 23(6, SI):1809–1821. Place: 233 SPRING ST,

NEW YORK, NY 10013 USA Publisher: SPRINGER

Type: Article.

Ghosh, C., Majumder, S., Ray, S., Datta, S., Mandal, and

S.N. (2020). Different EDGE Detection Techniques:

A Review. Lecture Notes in Electrical Engineer-

ing, 686:885–898. ISBN: 9789811570308 Publisher:

Springer Science and Business Media Deutschland

GmbH.

Gonzalez, C., Melin, P., and Castillo, O. (2017a). Edge

detection method based on general type-2 fuzzy logic

applied to color images. Information (Switzerland),

8(3). Publisher: MDPI AG.

Gonzalez, C., Melin, P., Castro, J., and Castillo, O. (2017b).

Edge detection methods based on generalized type-

2 fuzzy logic systems. SpringerBriefs in Applied

Sciences and Technology, 1(9783319539935):21–35.

Publisher: Springer Verlag.

Isa, M., Rohani, M., Ismail, B., Rosmi, A., Isa, M., Mustafa,

W., Kamarol, M., Mansor, N., and Nazar, N. (2021).

Review of Edge-based Image Segmentation on Elec-

trical Tree Classiﬁcation in Cross-linked Polyethylene

(XLPE) Insulation. In Journal of Physics: Confer-

ence Series, volume 1878. IOP Publishing Ltd. ISSN:

17426588 Issue: 1.

Jing, J., Liu, S., Wang, G., Zhang, W., and Sun, C. (2022).

Recent advances on image edge detection: A compre-

hensive review. Neurocomputing, 503:259–271. Pub-

lisher: Elsevier B.V.

Kaur, S. and Kaur, P. (2017). An Edge detection tech-

nique with image segmentation using Ant Colony Op-

timization: A review. In Proceedings of 2016 Online

International Conference on Green Engineering and

Technologies, IC-GET 2016. Institute of Electrical and

Electronics Engineers Inc.

Kitchenham, B. and Charters, S. (2007). Guidelines for per-

forming systematic literature reviews in software en-

gineering.

Koschan, A. and Abidi, M. (2005). Detection and classiﬁ-

cation of edges in color images. IEEE Signal Process.

Mag., 22(1):64–73.

Kumar, S., Singh, A., and Kumar, M. (2019). Information

hiding with adaptive steganography based on novel

fuzzy edge identiﬁcation. DEFENCE TECHNOL-

OGY, 15(2):162–169. Place: PO BOX 211, 1000 AE

AMSTERDAM, NETHERLANDS Publisher: ELSE-

VIER SCIENCE BV Type: Article.

Kumawat, A. and Panda, S. (2021). A robust edge de-

tection algorithm based on feature-based image reg-

istration (FBIR) using improved canny with fuzzy

logic (ICWFL). VISUAL COMPUTER. Place: ONE

NEW YORK PLAZA, SUITE 4600, NEW YORK,

NY, UNITED STATES Publisher: SPRINGER Type:

Article; Early Access.

Lopez-Molina, C., Bustince, H., Fern

andez, J., Couto, P.,

and De Baets, B. (2010). A gravitational approach

to edge detection based on triangular norms. Pattern

Recognition, 43(11):3730–3741.

Magnier, B. (2018). Edge detection: a review of dissim-

ilarity evaluations and a proposed normalized mea-

sure. MULTIMEDIA TOOLS AND APPLICATIONS,

77(8):9489–9533. Place: VAN GODEWIJCK-

STRAAT 30, 3311 GZ DORDRECHT, NETHER-

LANDS Publisher: SPRINGER Type: Review.

Magnier, B., Abdulrahman, H., and Montesinos, P. (2018).

A review of supervised edge detection evaluation

methods and an objective comparison of ﬁltering gra-

Fuzzy Logic Based Edge Detection Methods: A Systematic Literature Review

393

dient computations using hysteresis thresholds. Jour-

nal of Imaging, 4(6). Publisher: MDPI Multidisci-

plinary Digital Publishing Institute.

Marco-Detchart, C., Bustince, H., Fern

andez, J., Mesiar,

R., Lafuente, J., Barrenechea, E., and Pintor, J. M.

(2021a). Ordered directional monotonicity in the con-

struction of edge detectors. Fuzzy Sets and Systems,

421:111–132.

Marco-Detchart, C., Lucca, G., Lopez-Molina, C.,

De Miguel, L., Pereira Dimuro, G., and Bustince, H.

(2021b). Neuro-inspired edge feature fusion using

Choquet integrals. Information Sciences, 581:740–

754.

Marr, D. and Hildreth, E. (1980). Theory of edge detection.

Proceedings of the Royal Society of London. Series B.

Biological Sciences, 207(1167):187–217.

Martin, D. R. (2002). An empirical approach to grouping

and segmentation. University of California, Berkeley.

Mohan, K., Seal, A., Krejcar, O., and Yazidi, A. (2021).

Facial Expression Recognition Using Local Grav-

itational Force Descriptor-Based Deep Convolution

Neural Networks. IEEE Transactions on Instrumen-

tation and Measurement, 70:1–12.

Moya-Albor, E., Ponce, H., and Brieva, J. (2017). An

Edge Detection Method using a Fuzzy Ensemble Ap-

proach. ACTA POLYTECHNICA HUNGARICA, 14(3,

SI):149–168. Place: BECSI UT 96-B, BUDAPEST,

H-1034, HUNGARY Publisher: BUDAPEST TECH

Type: Article.

Mubashar, M., Khan, N., Sajid, A., Javed, M., and Has-

san, N. (2022). Have We Solved Edge Detection? A

Review of State-of-the-art Datasets and DNN based

Techniques. IEEE Access, pages 1–1. Publisher: In-

stitute of Electrical and Electronics Engineers Inc.

Muthalagu, R., Bolimera, A., and Kalaichelvi, V. (2020).

Lane detection technique based on perspective trans-

formation and histogram analysis for self-driving cars.

Computers & Electrical Engineering, 85:106653.

Orhei, C., Vert, S., and Vasiu, R. (2020). A Novel Edge

Detection Operator for Identifying Buildings in Aug-

mented Reality Applications. In Lopata, A., Butkien

R., Gudonien

e, D., and Sukack

e, V., editors, Informa-

tion and Software Technologies, volume 1283, pages

208–219. Springer International Publishing, Cham.

Pavithra, L. and Sharmila, T. S. (2018). An efﬁcient

framework for image retrieval using color, texture and

edge features. Computers & Electrical Engineering,

70:580–593.

Prewitt, J. M. et al. (1970). Object enhancement and

extraction. Picture processing and Psychopictorics,

10(1):15–19.

Qiu, B., Guo, J., Kraeima, J., Glas, H. H., Zhang, W., Borra,

R. J. H., Witjes, M. J. H., and van Ooijen, P. M. A.

(2021). Recurrent Convolutional Neural Networks for

3D Mandible Segmentation in Computed Tomogra-

phy. Journal of Personalized Medicine, 11(6):492.

Raheja, S. and Kumar, A. (2021). Edge detection based on

type-1 fuzzy logic and guided smoothening. Evolving

Systems, 12(2):447–462. Publisher: Springer Science

and Business Media Deutschland GmbH.

Roberts, J. (1980). Machine Percept 3D Solids.

Dissertations-G.

Russo, F. (1998). Edge detection in noisy images using

fuzzy reasoning. IEEE Transactions on Instrumenta-

tion and Measurement, 47(5):1102–1105.

Shimada, T., Sakaida, F., Kawamura, H., and Okumura, T.

(2005). Application of an edge detection method to

satellite images for distinguishing sea surface temper-

ature fronts near the Japanese coast. Remote Sensing

of Environment, 98(1):21–34.

Sobel, I., Feldman, G., et al. (1968). A 3x3 isotropic gradi-

ent operator for image processing. a talk at the Stan-

ford Artiﬁcial Project in, pages 271–272.

Sun, T.-H., Lai, C.-H., Wong, S.-K., and Wang, Y.-S.

(2019). Adversarial Colorization Of Icons Based On

Structure And Color Conditions. MM 19: Proceed-

ings of the 27th ACM International Conference on

Multimedia.

Suresh, K. and Srinivasa Rao, P. (2019). Various im-

age segmentation algorithms: A survey. Smart In-

novation, Systems and Technologies, 105:233–239.

ISBN: 9789811319266 Publisher: Springer Science

and Business Media Deutschland GmbH.

Tacconelli, E. (2010). Crd’s guidance for undertaking re-

views in health care. Lancet Infectious Diseases -

LANCET INFECT DIS, 10:226–226.

Tripathi, N., Kumrawat, D., Gottimukkala, V.K., Jeevaraj,

S., Godfrey, and W.W. (2021). Edge Detection on

Medical Images Using Intuitionistic Fuzzy Logic.

Communications in Computer and Information Sci-

ence, 1376 CCIS:236–246. ISBN: 9789811610851

Publisher: Springer Science and Business Media

Deutschland GmbH.

Wei, Y., Liang, X., Chen, Y., Shen, X., Cheng, M.-M., Feng,

J., Zhao, Y., and Yan, S. (2017). STC: A Simple to

Complex Framework for Weakly-Supervised Seman-

tic Segmentation. IEEE Transactions on Pattern Anal-

ysis and Machine Intelligence, 39(11):2314–2320.

Yadav, R. and Pandey, M. (2022). Image Segmentation

Techniques: A Survey. Lecture Notes on Data Engi-

neering and Communications Technologies, 90:231–

239. Publisher: Springer Science and Business Media

Deutschland GmbH.

Yang, M.-H., Kriegman, D., and Ahuja, N. (2002). Detect-

ing faces in images: a survey. IEEE Transactions on

Pattern Analysis and Machine Intelligence, 24(1):34–

58.

Yogesh, Dubey, A., Arora, R., and Ahmed, A. (2018). A

comparative review of various segmentation methods

and its application. In Khatri S.K., Kapur P.K., S. B.,

editor, 2017 6th International Conference on Reliabil-

ity, Infocom Technologies and Optimization: Trends

and Future Directions, ICRITO 2017, volume 2018-

January, pages 645–650. Institute of Electrical and

Electronics Engineers Inc.

Yuksel, M. and Basturk, A. (2012). Application of Type-2

Fuzzy Logic Filtering to Reduce Noise in Color Im-

ages. IEEE Comput. Intell. Mag., 7(3):25–35.

Zadeh, L. (1965). Fuzzy sets. Information and Control,

8(3):338–353.

Zhu, Y. and Li, P. (2019). Survey on the image segmen-

tation algorithms. Springer Series in Geomechanics

and Geoengineering, 1(216039):475–488. Publisher:

Springer Verlag.

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

394