You are on page 1of 6

2017 14th IAPR International Conference on Document Analysis and Recognition

A Comprehensive Survey on Handwriting and Computerized Graphology


Afnan H. Garoot Maedeh Safar Ching Y. Suen
Department of Computer Science and Department of Computer Science and Department of Computer Science and
Software Engineering Software Engineering Software Engineering
Concordia University Concordia University Concordia University
Montreal, Canada Montreal, Canada Montreal, Canada
af_garou@encs.concordia.ca ma_safar@encs.concordia.ca suen@cse.concordia.ca

Abstract— Graphology is a technique used to assess the As mentioned in [1], graphology is not a new research area,
writer’s personality traits from his/her handwriting features. its origins date back to the late 1900s and early 20th century.
Manual feature extraction and analysis is a time consuming Researchers started publishing books on handwriting analysis in
and labor intensive task. Therefore, computerized 1552, in Spain. More than 100 universal traits and symbols in
graphology systems have been developed by researchers to handwriting have been assembled by researchers in the last
overcome these issues. In this paper, we present the latest century.
state-of-the-art on computerized graphology systems.
There are many uses and applications of handwriting
Keywords— handwriting; graphology; recognition; feature analysis. The followings are the most popular applications used
extraction; personality trait; handwriting analysis; computerized today: dating and socializing, roommates and landlords,
graphology system. entertainment at parties and conventions, business and
professional, employee hiring and human resources, police
I. INTRODUCTION
profiling, self-improvements and professional speakers,
The study of a person’s character and personality from counsellors, therapists, and coaching applications [1].
his/her handwriting is known as Graphology or Handwriting
Baggett [1] explained in his book how handwriting reveals
Analysis. Graphologists use handwriting as an indication of
personality. He said that handwriting is often referred to as
personality trait represented by neurological patterns in the brain.
‘brain writing’. Neuromuscular movement tendencies have been
It is done first by extracting some specific features from the
categorized by research scientists in the neuroscience field as
handwriting samples. Then the extracted features are interpreted
they are related to specific observable personality traits. Each
based on some graphological rules. Handwriting analysis done
personality trait is represented by a neurological brain pattern,
by a computer helps graphologists to reveal and identify
which produces a unique neuromuscular movement that is the
personality of a person in an automated way. Development of
same for every person who has that personality trait. These tiny
computerized graphology systems is an active research area.
movements occur unconsciously when we write. Each written
Therefore, the goal of this paper is to present a literature review
movement or stroke reveals a specific personality trait.
on their methodologies and achievements.
Graphologists analyze these strokes as they appear in
The paper is organized as follows: section II describes handwriting and describe the corresponding personality trait.
graphology and handwriting features used in handwriting
Ahmed and Mathkour [3] stated that there are some
analysis. Section III explains the problems associated with
handwriting features which are used by graphologists to assess
extracting and analyzing features manually and describes
the personality traits. They used three kinds of measurements:
computerized graphology system and its advantages. Section IV
presents a state of the art on works related to computerized 1) General Measurements, which give an overall impression
graphology systems. Section V discusses the most recent papers. of handwriting. Graphologists associate personality traits with
Section VI summarizes our findings and presents future work. the stroke quality. For example, a small stroke in superior
handwriting is a measure of good concentration, while in inferior
II. GRAPHOLOGY
handwriting it measures pettiness.
In [1], Graphology is considered as a modern form of
2) Fundamental Measurements, which produce a primary
psychology that reveals personality traits, including emotional
classification of handwriting patterns. For this purpose, eight
outlay, fears, honesty, defenses and others, from the individual’s
basic features are used. These features are: slant which defines
handwriting, but not identifying the writer’s age, race, gender,
emotions, baseline direction which indicates the writer’s mood,
religion, or nationality. In other words, it is a technique used to
letter size which indicates power, continuity which indicates
evaluate and interpret the character of the writer from his/her
mode, handwriting form (shape) which indicates natural impulse
handwriting [2]. Nowadays, there are different formal tests on
and free choice, arrangement of lines on a page or words in lines
the market used to discover someone’s personality such as
which indicate sense of organization and adaptability, pen
aptitude studies, psychometric tests, and others. However, by
pressure which signifies the intensity, strength, and appetites,
comparing the practicality of these different diagnostic methods,
and writing speed which depicts the rhythm of physical and
it has been found that graphology is the fastest way.

2379-2140/17 $31.00 © 2017 IEEE 621


DOI 10.1109/ICDAR.2017.107
mental activity of the writers. IV. RELATED WORKS ON COMPUTERIZED
GRAPHOLOGY SYSTEMS
3) Accessories Measurements, which are graphic symbols
that appear in ‘t’ bars, ‘i’ dots, punctuation marks, capital letters, Early research on computerized graphology system was done
signatures, numerals, initial and terminal strokes, covering in 1995 by Sheikholeslami et al. [4]. They developed a system
strokes, flourishes, upper and lower extensions or loops and called Computer Aided Graphology (CAG), which consists of
alphabet. The letter ‘t’ is the most important graphical symbol four modules, which are: scanning, preprocessing, feature
that signifies the writer’s will power and activeness. The letter ‘i’ extraction, and feature analysis. They performed three
is the second most important graphic symbol which signifies the preprocessing operations on scanned images, which are:
speed, sphere of the imagination, intellect, ideals and aspiration. removing digitization artifacts, thresholding using Otsu method
In addition, the size of a capital letter represents the pride, vanity [5], and guide line removal using Hough Transform algorithm
and desire to impress others. The beginning strokes signify the [6]. They segmented the input image into lines in order to extract
consciousness, immaturity, conventionality, love of gain, features. Based on information collected from line images, the
selfishness, gaiety, while the ending strokes signify practicality, following features were extracted: page left margin, page right
aspirations, and courage of the writer. Moreover, the loops margin, page bottom margin, line spacing, line direction, slant,
occurring in letters also contribute personality related and upper, middle, and lower zone ratios. The system utilized
information like intellectual and physical activities, moderation, syntactic pattern recognition to interpret features. The developers
exaggeration, fluency of thoughts, idealism, aspirations, implemented a context free grammar that includes about 50 rules
ambition and spirituality, little vision of imaginations, and to map features to interpretations. They used lex and yacc
realistic practical viewpoints, egoism, boastfulness, vanity, software in lexical processing to generate the output code based
timidity and inhibition, fear and apprehension for future etc. on the specified grammar. They used 25 handwriting samples as
a test data set. As a result, most of the features had the same
III. COMPUTERIZED GRAPHOLOGICAL SYSTEM graphologist interpretations, but for some of them they were
In 1995, Sheikholeslami, Srihari and Govindaraju [4] mentioned different. In general, the output of the system was consistent
a number of problems associated with extracting features with graphologists' interpretation.
manually from handwriting samples. These problems are: (1) it
In 2004, an offline writer-independent handwriting analysis
is a tedious, subjective, and error prone task, (2) the same
system was presented [8]. They used classical crisp and fuzzy
features of the same handwriting sample might be extracted
methodologies in order to produce personality traits of the
differently by different graphologists, (3) the analysis of
writer. The system deployed an analytical handwriting analysis
graphologists might be influenced by the content of handwriting,
approach based on two primitives, the baseline and the slant
and (4) validity of graphology rules used in analyzing features is
angle of the characters. A non-fuzzy (crisp) paradigm was used
still in question. Therefore, a computerized graphology system
to evaluate the baseline while a fuzzy paradigm was used to
has been developed in order to overcome these problems.
evaluate the slant-angle. They identified 15 personality traits and
A computerized graphology system helps graphologists to created sets of rules.
extract handwriting features and analyze them faster and more A few years later, personality analysis based on small letter
precisely using computers. It takes a handwriting sample as an ‘t’ was proposed by Mutalib et al. [9] in offline handwriting. The
input and produces a personality description of the writer as an first step was recognition of small letter ‘t’ character from a
output. It usually consists of four main modules which are: 1) scanned image of documents once several methods of
scanning, 2) preprocessing, 3) feature extractions, and 4) preprocessing applied, then analyzing three different angles of t-
analyzing features. [7] mentioned a number of advantages of a bars (up-turned, horizontal and down-turned) which would help
computerized handwriting analysis, which are: fast, accurate, to determine the level of writers’ ambition at the moment of their
identifies the handwriting better than visual assessment, writing. This study used two back propagation neural networks
efficient, and devoid of human errors. It is being applied in to recognize small ‘t’ character and to classify the level of
different areas such as business, psychiatry, medicine and ambition.
criminology [3].
In [3], the construction of an automated graphology system
Previously, various researchers have studied handwriting was presented. The system deployed a rule-based method which
analysis with the aid of a computer. Several reviews are focused was equipped with the conventional pattern recognition
on existing technologies and implementation of different techniques and an inference engine to make decisions along with
methodologies in graphology [17], various types of features [22] an explanation. The architecture of the rule-based system
and applications of personality assessments [23]. However, this consists of three modules: feature extraction, inference engine,
paper reviews related works on computerized graphology and knowledgebase. Eight features were used by the system,
systematically in terms of preprocessing operations, extracted which are: slant, base line, speed, size, continuity, form,
features, analysis techniques, used data bases and experimental arrangement, and pressure. However, six personality attributes
results. were computed by the system, which are: emotions, mood, self-
confidence, coherence of thoughts, strength and organization. 35

622
students’ handwriting samples were used by the system as a test which focused particularly on the two important features to
data set. Since the system presented in this study was in a analyze and predict the writer’s personality in offline system.
development stage, the results needed to be more accurate in Baseline and writing pressure were the features considered by
measurements, reliability, and the features needed to be them in this article. To categorize personalities, they applied a
extracted in a finer manner. rule-based approach with 9 rules formed by three types of
baseline (ascending, descending and level), and also three kinds
In the following year, graphology has been used in
of writing pressure (light, medium and heavy).
recruitment process in many human resource companies. In [10],
aptitudes of a job applicant were measured by developing a In [7], an automated handwriting analysis method was
model that links features from handwritten images to a number proposed to predict a person's behavior. The handwriting was
of personality traits. They extracted the following features: the analyzed using image processing MATLAB. It consists of three
slant, the slope, the margins, the line spacing, the size of the steps: image processing, feature extraction, and prediction. The
script, angular writing and regularity in the writing frequency, RGB (Red, Green and Blue) threshold was performed in order to
the pen pressure, image contours and fractal dimension. After isolate the ink image from the background. Eight characteristic
feature extraction process completed, data was classified using a traits of handwriting were determined: size of the letters,
neural network. They took 98 samples from people of different baseline, pressure of the writing, slant of the writing, breaks in
ages and languages. As a result, the classification got a the writing, spacing between the words, margins, and speed of
successful rate of 89%. the writing. The handwriting samples were taken from 30 people
of different ages. They were analyzed manually by a handwriting
In 2010, Champa and Kumar [11] proposed a method to
analyst and the inferences were made. The features determined
recognize the trait of a person based on the baseline, the pen
by the system were compared against a reference. The results
pressure, and the letter ‘t’. They used polygonalization method in
showed a good agreement with more than 80% of the cases
order to evaluate the baseline and grey-level threshold value in
based on ideal manual analysis.
order to evaluate the pen pressure. For the position of the t-bar,
five predefined templates were specified to compare and match. In the following year (2012), Grewal and Prashar [16]
In the same year, Champa and Ananda Kumar [12] improved worked on features like baseline, the letter slant, pen pressure
their previous work and added several features such as the lower same as the other researchers as the main features which can
loop of letter ‘y’ and the slant of the writing to review and contain accurate information about the personality trait.
compare the results. The form of the lower part of letter ‘y’ was Furthermore, they did research on 6 common formations of the
calculated by the Generalized Hough Transform algorithm. letter ‘i’ and the letter ‘f’ which show the ability of planning and
Template matching as a character recognition method and organization of writer. After feature extraction, all these features
hamming distance as a decision making technique are deployed were used as inputs of ANN to predict personality of the writer.
for obtaining the slant of writing. They used a database that
includes 120 handwriting samples from 120 writers who were At the same time, another study [17] examined existing
not the same. Their results showed high efficiency and technology that can identify or predict personality traits. In
consistency between manual and automated method. addition, they examined some technologies that identify and
authenticate handedness, authorship, gender using handwriting
Another research was conducted in 2010 [13]. It was about samples. They also referenced several implementations of the
developing an automated method to predict the personal different methodologies/technologies and techniques. They
behavior of individuals from the digital form of their found that all technologies which have been discussed do not
handwriting. They considered six main types of features: size of have a complete vision and none of the tools can accurately
letters, slant of letters and words, baseline, pen pressure, spacing predict a deviant behavior.
between letters, and spacing between words. They used
segmentation to calculate the features from digital handwriting. In 2013, according to [18] another system was proposed to
Once the features were extracted, they were trained with Support predict the personal behavior of individuals from their digital
Vector Machine (SVM) which outputted the behavior of the handwriting samples. The behavioral analysis was done based on
writer. Handwriting image samples were collected from 100 the following features: the pen pressure, the slant of letters and
different writers, and then they were digitized using the scanner. the slant of baseline, the size of letters, spacing between words.
The proposed method gave about 94% of accuracy rate with a To implement the system, simple linear regression method was
Radial Basis Function (RBF) kernel. This research has become used, which is an approach to model the relationship between a
more complete in 2014 by Raut and Bobade [14]. Margins and scalar dependent variable y and an explanatory variable denoted
speed of the writing were added to the six features mentioned by x. The implemented method proved successful in analyzing
above and all these eight parameters were given to the SVM as handwriting regardless of the language used.
an input. A variety of kernels such as linear and polynomial were Another automated system was presented in 2013 by Djamal
tested with SVM; however, the RBF produced a better accuracy, and Darmawati [19] who employed five structural features,
near 90%. namely, margins, spacing between lines and words, domination
In 2011, Champa and Kumar [15] published another paper zone and baseline. This system used multi-structure algorithms

623
for classification of four features except baseline which resulted which are applied for personality identification through
in 97-100% accuracy rate, and utilized hill valley and artificial handwriting analysis. The principal part of this review is on
neural networks for baseline which did not lead to an impressive various features which are extracted in related works. Besides,
accuracy (78%). they mentioned steps of the procedure of identifying like
preprocessing, scanning, feature extraction and classification in
Several research studies have been done in 2015. Based on
brief. Kedar and Bormane [23] did another review on automatic
[20], a tool for predicting the personality traits from Farsi
personality assessment. It provides information about
handwriting was developed. The preprocessing steps included:
applications of personality assessment, various techniques used
pen width extraction, noise and scratch removal using a
for automatic assessment of personality, and guidelines for
morphological closing operation followed by a morphological
future research. It has been observed that the psychology tests
opening operation. After the completion of preprocessing
available for personality assessment usually include a long
operations, seven features were extracted: right and left page
questionnaire which is very time consuming and tedious. This
margins, word expansion, letter size, line and word spacing, line
leads a person to lie in order to create a fake and impressive
spacing, word spacing, line skew, the ratio of vertical to
personality. However, the online tools used for automatic
horizontal elongation of words, and slant. They used Support
personality assessment using handwriting analysis require basic
Vector Machine (SVM) classifier to predict the personality. The
knowledge of various writing features. Therefore, the person
input database included 120 handwriting samples from 120
with this knowledge can use these tools efficiently and
different writers. The system showed promising results. Another
effectively.
study was done by Gavrilescu [21] to analyze the link between
personality types and handwriting, through correlating the
The recent research in 2016, regarding automated graphology
handwriting features with the personality primitives in a neural
system reported a method developed for segmentation, baseline
network 3-level architecture. The following features were
recognition and pressure of the writing. Bal and Saha [24]
extracted: baseline, writing pressure, word slant, connecting
improved techniques of horizontal and vertical projections to
strokes, lowercase letter "t", and lowercase letter "f". The
decline incorrect line segmentation due to overlapping. Thus
database contained 64 subjects and the results showed an
Lines and words segmentation and also skew normalization
accuracy of 86.7% in determining the personality type, with
methods were developed in this research. Furthermore, the
highest accuracies for Extravert vs. Introvert and Thinking vs.
proposed methods are able to predict personality of the writer
Feeling personality primitives.In the same year (2015), two
through baseline and pressure of the writing.
surveys were done by a group of researchers. In the first one,
Kedar, Nair and Kulkarni [22] reviewed different methodologies

TABLE I. RELATED WORKS ON COMPUTERIZED GRAPHOLOGY SYSTEM


Analysis
Authors and Year Preprocessing Operations Extracted Features Data Base Results
Technique
The output of the
Removing digitization
Sheikholeslami et Margins, line spacing, line Syntactic pattern 25 handwriting system was consistent
artifacts, thresholding, and
al. (1995) direction, slant, and zone ratios. recognition. samples with graphologists'
guide line removal
interpretation.
Classical crisp and
Mogharreban et al.
Not mentioned The baseline and the slant. fuzzy Not mentioned Not mentioned
(2004)
methodologies
Digitization and Average accuracy for
normalization, scaling, Two back the small letter ‘t’
Mutalib et al. 50 respondents to
noise removal, binary Small letter ‘t’ character. propagation neural recognition is 90.27%
(2007) questionnaires
conversion and networks and for the ambition
segmentation recognition is 60%.
A rule-based system
Slants, base line, speed, size, equipped with the 35 students’ Not mentioned, but in
Ahmed &
Not mentioned continuity, form, arrangement, and conventional handwriting general the results were
Mathkour (2008)
pressure techniques and samples encouraging.
inference engine
Slant, slope, line spacing, margins,
size, image contours, angular 98 samples from
Back propagation A correct classification
Coll et al. (2009) Not mentioned writing, fractal dimension, people of
neural network rate of 89%.
regularity in the writing frequency, different ages.
pen pressure
Template matching,
Champa & Ananda Baseline, the pen pressure, and the
Polygonalization, and ANN Not mentioned Not clear
Kumar (2010) letter ‘t’.
gray level thresholding.
A database that High efficiency and
Template matching, Baseline, the pen pressure, the letter
Champa & Ananda includes 120 consistency between
Polygonalization, and ‘t’, the lower loop of letter ‘y’ and Rule-Based system
Kumar (2010) handwriting manual and automated
gray level thresholding. the slant of the writing.
samples method.

624
Authors and Year Preprocessing Operations Extracted Features Analysis Technique Data Base Results
Size, slant, baseline, pen pressure, Support Vector 100 handwriting 94% of accuracy rate
Prasad et al. (2010) Not mentioned
spacing between letters and words. Machine (SVM) samples with RBF kernel
Polygonalization,
Champa& Ananda converting scanned image Not mentioned
Kumar (2011) into a binary image by Baseline and pen pressure Rule-Based system Multiple samples
using a thresholding
algorithm.
The handwriting
The results were in
samples were
Size, baseline, writing pressure, good agreement to
analyzed manually
slant, breaks in the writing, spacing 30 handwriting more than 80 percent of
Kamath et al. RGB threshold by a handwriting
between the words, margins, and samples the cases with ideal
(2011) analyst and the
speed of the writing. manual analysis.
inferences were
made.
Mean Square Error
Grewal and Template matching,
Baseline, slant, pen pressure, letter (MSE) reduces as
Prashar polygonalization and ANN 50 handwritings.
‘i’ and letter ‘f’ number of epochs
(2012) gray level thresholding
increased.
Fisher et al.
A limited survey on existing technologies and implementation of different methodologies in graphology.
(2012)
Proved successful in
The baseline slant, the pen pressure,
Resizing to the correct Simple linear analyzing handwriting
Abdul Rahiman et the slant of the writing, size of Not mentioned
orientation regression method regardless of the
al. (2013) letters, and spacing between words
language used.
97-100% accuracy rate
Multi-structure 25 Handwritings for classification of
Djamal & Margins, spacing between lines and
Grayscale, thresholding, algorithms and for training data. personality. 78%
Darmawati words, domination zone and
normalization and cropping artificial neural 38 Handwritings accuracy rate for the
(2013) baseline
networks for testing data classification of
baseline.
Smoothing by removing the Pen pressure, baseline, size of the
Using the RBF shows a
Raut & Bobade noise, line segmentation, letters, spacing between letters and Support Vector 100 handwriting
better accuracy, near
(2014) letter segmentation and words, slant, margins and speed of Machine (SVM) samples
90%.
word segmentation the writing
Margins, word expansion, size, line Not mentioned, but in
Hashemi et al. Pen width extraction, noise and word spacing, line skew, the Support Vector 120 handwriting general the system
(2015) and scratch removal ratio of vertical to horizontal Machine (SVM) samples showed promising
elongation of words, and slant. results.
Normalization,
removing noise, isolating Baseline, Writing pressure, slant,
A neural network 3- Results showed an
Gavrilescu (2015) handwritten text, and Connecting strokes, Lowercase 64 subjects
level architecture. accuracy of 86.7%
converting image to letter "t" and "f".
grey scale
Kedar, Nair and A review paper on different methodologies which are applied for personality identification through handwriting analysis. The principal part
Kulkarni (2015) of this review is on various features which are extracted in related works.
Kedar and A review paper which provides information about applications of personality assessment and various techniques used for automatic
Bormane (2015) assessment of personality.
Noise removal, IAM database Accuracy rate of lines
thresholding, line over 550 text segmentation and
segmentation, skew images words segmentation are
Bal and Saha normalization, measures containing 3800 95.65% and 92.56%,
Baseline and writing pressure Rule-Based system
(2016) width of inter-word and words and some respectively. 96% of
intra-word gaps, normalize sample lines and words were
the skew of the segmented handwriting normalized perfectly
words image with tiny error rate.

V. DISCUSSION order to identify the letter size, another index can be calculated
by finding the gravity centers of the connected components [20].
Based on the recent works, different methods can be used to The line spacing can be found by dividing the number of black
extract handwriting features. For page margin, the side pixels of pixels by the product of the area of the convex hull and the pen
the dilated end row points of the image can be extracted and width [20]. To find the skew, the angle corresponding to
considered as margins [20]. For the word expansion, a specific minimum entropy of the horizontal projection can be
index can be used and determined by labeling the connected determined. For calculating the slant, different methods can be
components and dividing their area by the pen width [20]. In used such as a polygonization technique and other algorithms

625
based on contour extraction and chain code implementation [20] of Engineering and Applied Sciences, 6(9), pp.1819-660.
[21]. For extracting the baseline, a polygonization technique can [8] N. Mogharreban, S. Rahimi, and M. Sabharwal. A combined crisp and fuzzy
be adopted [21]. Writing pressure can be determined using the approach for handwriting analysis. In Fuzzy Information, Processing NAFIPS'04.
IEEE Annual Meeting of the, vol. 1, pp. 351-356, 2004.
grey level thresholding algorithm and the standard deviation [21]
[9] S. Mutalib, S. Abdul Rahman and M. Yusoff. Personality analysis based on
[24]. For the lower case letter “t” and “f”, template matching can letter ‘t’ Using back propagation neural network. In Proceedings of the
be used. International Conference on Electrical Engineering and Informatics Institut
Teknologi Bandung, Indonesia June (pp. 17-19), 2007.
VI. CONCLUSION [10] R. Coll, A. Fornés, and J. Lladós. Graphological analysis of handwritten text
documents for human resources recruitment. In Document Analysis and
Recognition, 2009. ICDAR'09. 10th International Conference on, pp. 1081-1085.
Handwriting consists of a bunch of written symbols which IEEE, 2009.
can be considered as brain writing. More specifically, the [11] H. Champa and K. Ananda Kumar. Automated human behavior prediction
handwriting is an indicative tool to reveal the personality of the through handwriting analysis. In Integrated Intelligent Computing (ICIIC), 2010
writer [25]. According to prior studies we have identified several First International Conference on Integrated Intelligent Computing, pp. 160-165.
IEEE, 2010.
important features that have been used in computerized
[12] H.N. Champa and K. R. Ananda Kumar. Artificial neural network for
graphology such as page margin, baseline, line and word human behavior prediction through handwriting analysis. International Journal
spacing, line direction, slant of words and letters, size of letters, of Computer Applications (0975–8887) Volume. 2010.
pen pressure, speed of the writing, the letter ‘t’, letter ‘i’, letter [13] S. Prasad, V. K. Singh and A. Sapre. Handwriting analysis based on
‘f’, the lower loop of letter ‘y’, etc. In addition, it has been segmentation method for prediction of human personality using support vector
observed that different analysis methods such as support vector machine. International Journal of Computer Applications 8(12), pp. 25-29. 2010.
machine (SVM), artificial neural networks (ANN), rule-based [14] Anike A. Raut, Ankur M. Bobade, “Prediction of human personality by
handwriting analysis based on segmentation method using support vector
systems have been used widely in graphology. However, there is machine”, International journal of Pure and Applied Research in Engineering and
a powerful analysis method, called deep learning, has not been Technology, May 2014
applied yet in this area. Therefore, applying deep learning [15] H. N. Champa, and K. R. Ananda Kumar. "Rule Based Approach for
method for designing a fully automated graphology system with Personality Prediction Through Handwriting Analysis." In 2nd International
a high accuracy rate should be considered as a future work. The Conference on Biomedical Informatics and Signal processing, organized by Saiࣔ
s BioSciences Research Institute Pvt. Ltd. 2009.
current trend of computerized graphology shows that applying a
[16]P.K. Grewal & D. Prashar."Behavior prediction through handwriting
combination of analysis methods [19] and using features of analysis." IJCST 3, no. 2 (2012).
signature besides handwriting features [19] resulted in [17] J. Fisher, A. Maredia, A. Nixon, N. Williams and J. Leet. Identifying
remarkable accuracy. Moreover, the quantity of training data has personality traits, and especially traits resulting in violent behavior through
an impact on achieving better results [14]. automatic handwriting analysis. Proceedings of Student-Faculty Research Day,
CSIS, Pace University,2012.
ACKNOWLEDGMENT [18] A. Rahiman, D. Varghese and M. Kumar. Habit: Handwritten analysis based
individualistic traits prediction. International Journal of Image Processing (IJIP)
7(2), pp. 209. 2013.
This research was supported by a research grant from [19] E. C. Djamal & R. Darmawati."Recognition of Human Personality Trait
NSERC, Natural Sciences and Engineering Research Council of Based on Features of Handwriting Analysis Using Multi Structural Algorithm
Canada. and Artificial Neural Networks." In 2013 IEEE Conference on Control, Systems
& Industrial Informatics (ICCSII), pp. 22-24. 2013.
REFERENCES [20] S. Hashemi, B. Vaseghi, and F. Torgheh. Graphology for Farsi Handwriting
Using Image Processing Techniques, In IOSR Journal of Electronic and
Communication Engineering, Volume 10, pp.01-07, 2015.
[1] B. Baggett. Handwriting analysis 101-the basic traits. Published by Empresse
Publishing and Distributed by Handwriting University,2004. [21] M. Gavrilescu, Study on determining the Myers-Briggs personality type
based on individual's handwriting. In E-Health and Bioengineering Conference
[2] H. Ploog. Handwriting Psychology: Personality Reflected in
(EHB), 2015, pp. 1-6. IEEE, 2015.
Handwriting,2013.
[22] Kedar, Seema, Vaishnavi Nair, and Shweta Kulkarni. "Personality
[3] P. Ahmed and H. Mathkour. On The Development of an Automated
identification through handwriting analysis: a review." Int. J. Adv. Res. Comput.
Graphology System. In IC-AI, pp. 897-901. 2008.
Sci. Softw. Eng 5, no. 1 (2015).
[4] G. Sheikholeslami, S. Srihari, and V. Govindaraju. "Computer aided
[23] S. Kedar and D. Bormane. Automatic personality assessment: A systematic
graphology." Master's thesis, State University of New York at Buffalo, 1995.
review. In International Conference on Information Processing (ICIP), pp. 326-
[5] Nobuyuki Otsu. "A threshold selection method from gray-level 331. IEEE, 2015.
histograms." Automatica 11, no. 285-296 (1975): 23-27
[24] A. Bal and R. Saha. An Improved Method for Handwritten Document
[6] Hamarneh, Ghassan, Karin Althoff, and Rafeef Abu-Gharbieh. "Automatic Analysis Using Segmentation, Baseline Recognition and Writing Pressure
line detection." Project Report for the Computer Vision Course Lund, Simon Detection. Procedia Computer Science, 93, pp.403-415. 2016.
Fraser University (1999).
[25]C. P. Leibel. Change your handwriting, change your life. Barnes & Noble,
[7]Kamath, V., Ramaswamy, N., Karanth, P.N., Desai, V. and Kulkarni, S.M., 1973.
2011. Development of an automated handwriting analysis system. ARPN Journal

626

You might also like