2nd Int'l Conf on Electrical Engineering and Information & Communication Technology (ICEEICT) 2015

Jahangirnagar University, Dhaka-I 342, Bangladesh, 21-23 May 2015

Improving Fingerprint Based Gender Identification

Technique Using Systematic Pixel Counting
Md. Sazzad Hossain; Md. Ahsan Habib

Dept. of Computer Science and Engineering

Begum Rokeya University, Rangpur
Rangpur -5400, Bangladesh
[email protected]; [email protected]

Abstract-Finger prints are considered as the best tool of

identification.

Gender

identification

from

fingerprints

is

an

important step in forensic anthropology in order to identify the

gender of a criminal and minimize the list of suspect search.
Fingerprint identification and classification has been extensively
researched in the literature however very few researchers have
studied the fingerprint gender classification problem. In the
present work, we proposed an easy and efficient technique to
classify gender of a person using systematic fingerprint ridge
counting to a valley area. The study was carried out on the age
group of 18 to 40 years where the students, workers, housewives
etc are taken as the subj ects. An optimal threshold for each
transform is chosen for better results.

It is found that the

proposed system produces accurate decision of 85% of male and

74% of female. The overall performance of the system is found to

be satisfactory and useful to forensic anthropology.

Keywords-Fingerprint;, Gender; ridge counting; valley area;

threshold;

I.

different biometrics traits such as face, gait, iris, hand shape,

speech and fingerprint. Fingerprint has been used as a
biometric for the gender identification because of its unique
nature and do not change throughout the life of an individual.
Fingerprints have ridge characteristics that allow for efficient
classification and examination.
Generally ridge related parameters such as fingerprint
ridge count, ridge density and thickness to valley thickness
ration, ridge width and fingerprint patterns and the pattern
types are mostly used for gender determination.
In our work, all fingerprints are acquired from random
subjects. This fingerprint gender identification system
constitutes of digital images of fingerprint as its input which is
then going through some preprocessing steps to make the
image computable. Lastly We can compute the image to make
a decision. The algorithm can be developed in any
programming language. We choose C# .Net framework 4.0 to
implement this method.

INTRODUCTION

Nowadays, with security being so important, the field of

biometrics is becoming increasing. The skin is marked by
"carved works" especially on the palms and soles. The pattern
of these markings on the fingers is called Dermatoglyphics[1].
In Dermatoglyphics, the impressions made by the minute ridge
patterns have exactly the same arrangement and the pattern of
any individual remain unchanged throughout life. These
features statistically differ among sexes, ethnic groups and age
categories [2]. Gender identification of suspect from crime
scene is an important task in forensic science that minimizes
the search population of suspects. In addition, judges put the
higher value on such evidence like fingerprints than other
physical evidence [10]. The science of fingerprint has been
used generally for the identification or verification of person
and official documentation. Fingerprint analysis plays a
role in convicting the person responsible for an audacious
crime. Fingerprints are the most effective identification of
evidence. Especially in criminal identification, in case of
mlssmg person or disaster vIctIms, the fingerprint
identification is the first simple individual identification before
using complicated technique such as DNA analysis. Gender
information is important to provide investigative leads for
finding unknown persons. Fingerprints may be classified and
documented on the basis of ridge pattern, setting up an
identification system that is almost infallible [3]. Various
methodologies has been used to identify the gender using

978-1-4673-6676-2/15/$31.00 20151EEE

II.

RELATED WORK

Studies so far carried out to findings gender of a sample

are based on the spatial domain analysis of ridges. Nithin MD,
Manjunatha B, Preethi OS, Balaraj BM in 2011 presented a
study with a goal to determine the gender based on finger
ridge count within a well-defined area [4]. Rolled fingerprints
were taken from 550 subjects (275 men and 275 women)
belonging to South Indian population all within the age range
of 18-65 years. Results showed that women have a
significantly higher ridge count than men. Application of
Baye's theorem suggests that a fingerprint possessing ridge
density <13 ridges/25 mm2 is most likely to be of male origin.
Likewise, a fingerprint having ridge count>14 ridges/25 mm2
is most likely to be of female origin.
Ahmed Badawi, et ai, (June, 2006) proposed a method for
Gender classification from fingerprints. A dataset of 10fingerprint images for 2200 persons of different ages and
gender (1100 males and 1100 females) was analyzed. Features
extracted were; ridge count, ridge thickness to valley thickness
ratio (RTVTR), white lines count, and ridge count asymmetry,
and pattern type concordance. Fuzzy- C Means (FCM), Linear
Discriminate Analysis (LOA), and Neural Network (NN) were
used for the classification using the most dominant features
predicted results of 80.39%, 86.5%, and 88.5% using FCM,
LOA, and NN, respectively [5].

Murlidhar Reddy Sangam, Karumanchi Krupadanam,

Kolla Anasuya, in 2011, presented a study that revealed that
there is significant sex and bimanual differences in the
distribution of the finger print pattern. Their study was
undertaken to observe the distribution of finger print pattern in
males and females, and to observe the bilateral asymmetry in
the region of Andhra Pradesh [6]. Whorls were of high
frequency on thumb, index and ring fmgers in males. But
females presented high frequency of loops on all fingers
expect ring finger. There is a significant bimanual difference.
Whorls are more common in right hands. Arches and radial
loops are more on left index finger. The study revealed that
there were significant sex and bimanual differences in the
distribution of the finger print pattern.

III.

PREPARED TECHNIQUE

The proposed work has been divided into four sections

namely:

Finger Print Collection

Preprocessing

Feature Extraction

Decision Making

The overall method can be shown in Figure-2.

Jen feng wang, et ai, (2008) worked on gender

determination using finger tip features. He obtained
fingerprints from 115 normal healthy adults in which 57 were
male fingerprints and 58 were female fingerprints. They have
used ridge count, ridge density, and fmger size features for
classification. However, the ridge count and finger size
features of left little fmgers are used to achieve a
classification. The best classification result of 86% accuracy is
obtained by using ridge count and fmger size feature together
[7].
Ramanjit Kaur, Rakesh K. Garg in 2011 with their study
provided an aid for the fingerprint examiner in analyzing
fingerprint samples as it shows that there is a significant
difference in epidermal ridge density between males and
females of the two populations [8]. Their study has been
carried out to examine ridge density differences in two
Northern Indian populations (Sikh Jat and Bania). In their
study it has been found that 92% of Sikh Jat females have a
mean ridge density above 13, whereas 76% of Sikh Jat males
have (a mean ridge density) be low 13, while in Bania, 100%
of females have mean ridge density above 14 and 80% of
males - below 14. The study suggested that there are
significant differences in epidermal ridge density between
males and females within each of the two populations, and
also significant differences between the two populations.
Acree, M. in 1999 presented a study whose aim is to
determine if women have significantly higher ridge density,
hence finer epidermal ridge detail, than men by counting
ridges that occur within a well-defined space [9]. If significant
gender differences do exist then the likelihood of inferring
gender from given ridge densities will be explored. Their
study focused on 400 randomly picked ten-print cards
representing 400 subjects. The demographic composition of
this sample population represents 100 Caucasian males, 100
African American males, 100 Caucasian females and 100
African American females all within the age range of 18-67.
Results show that women tend to have a significantly higher
ridge density than men and that this trend is upheld in subjects
of both Caucasian and African American descent (F 81.96, P
< 0.001). Application of Bayes' theorem suggests that a given
fingerprint possessing a ridge density of 11 ridges/25 mm2 or
less is most likely to be of male origin. Likewise a fingerprint
having a ridge density of 12 ridges/25 mm2 or greater is most
likely to be of female origin, regardless of race.
=

Fig. 1. Example Proposed System

A. Fingerprint Collection

The fingerprint images of internal database were collected

from students, workers, housewives etc. The images were
captured with a resolution of 1024x768. For this work a
database with 400 images of left and right thumb fingerprints
was created which correspond to 100 Male and 100 Female.
Every subject of this database is having age between 18 to 40
years.
B. Preprocessing

After collecting fingerprint samples, some preprocessing

work have been carried out such as background elimination,
resize image, convert the fingerprint to binary image and
define the valley area. Primarily the input image is a color
image. For computational efficiency, the high intensity level
color image is converted to binary image. We define 200x200
pixels as valley area where the center point of the processed
image and center point of the valley area are same. Figure-2
shows an example of Image processing where Figure-2 (a)

shows a raw figure print image whether Figure-2 (b) shows

processed version of the first image. Here into the Figure-2 (b)
red marked area indicates the valley area.

111.

Examine all pixels under the valley area and count

number of white pixels as W and number of Black
pixels as B.

iv.

Calculate the ratio R between black and white pixels

by following (1)".
"

v.
vi.
Fig. 2. Prosessing of a sample Image
C.

vii.

Compare R with the threshold point Th.

If R>=Th, then declare the image from a Female
subject.
Otherwise, declare the image from a Male subject.

Feature Extraction

Feature extraction is a fundamental preprocessing step for

pattern recognition and machine learning problems. The
central problem is designing a fingerprint classification system
to determine what features should be used and how categories
are defined based on these feature. Ridges are defined as the
black impression in the fmger print images while valises are
defined by the white portion. In this section, we complete total
calculation by following three steps. At first, we count the
number of white pixels within the valley area. Secondly, we
count the number of black pixels within the same area. Finally
we calculate the ratio between black pixels and white pixels. If
we represent the number of white pixels as Wand the number
of black pixels as B then we can easily calculate the ratio R
between them as in
R= B / W
D.

(1)

Decision Making

Decision Making is the final step of this method. In this

step we compare the calculated ratio R with a standard
threshold point. An optimal threshold point set for each
transform is an important part of the gender identification
process. Here, to make the calculation successful we set the
threshold point very carefully based on experimental data
where 400 fingerprints of both male and female were
examined. After analyzing all fingerprints, we found an
efficient threshold point that we called Th and the value of
this point is
Th= 0.51

(2)

We set this value, because we show in our analysis, for any

value of R greater than or equal to 0.51 is more likely to a
female subject, otherwise the subject is more likely to a male.
IV.

ALGORITHM DESIGN

i.

Get cover image

ii.

Firstly, Convert Input Image to Binary Image. That

means set all pixels by only two possible values Black
or White. Secondly, define a valley area. Here
200x200 pixels of the input image from the center of
the input image are defined as valley area.

Fig. 3. Proposed Algorithm

EXPERIMENTAL RESULT

V.

Table I shows distribution of white and black pixels of

fingerprint samples according to Gender. The results of the
distribution show that the majority of the male subjects are
having black and white pixels ratio R less than 0.51 that is the
threshold point Th. On the other hand most of the female
subjects are having R greater than or equal to the threshold
point Th.
TABLE!.

RESULT OF PIXEL COUNTING

R>=S.l

R<O.Sl

Gender

Total

Female

52

26.00%

148

74.00%

200

Male
Total

171
223

85.50%

29
177

14.50%

200
400

VI.

CONCLUSION

In this research, we have discussed the technical approach

followed in processing input images, feature extraction and
analyzing processes of a finger print image. We have
presented a new system for the Fingerprint Based Gender
identification which can be ensured accurate decision of 85%
of male and 74% of female here. After studying of this system,
we can conclude that there was an association between
distribution of fingerprint images black and white pixels and
gender group of a person. We also can conclude that gender
classification is depended on qualities of finger print images
and maximum number of images should be taken as obtaining
good result or reaching high efficiency of result.
ACKNOWLEDGMENT

We are mostly benefited for receiving the pre-papers on

the very topic which are helpful for us to develop our idea. We
are grateful to those who guided us keeping up our project
deeds. Our heart humble thank goes to them, although it is not
possible to return what we have received from them within
this short time.
REFERENCES
[1]

G. M. Bhat, M. A. Mukhdoomi, B. A. Shah and M. S. Ittoo,

Dennatoglyphics: in health and disease - a review, International Journal
of Research in Medical Sciences 2(1),2014,31 -37.

[2]

S. S. Gornale, C. D. Geetha, and R. Kruth, Analysis of Fingerprint

Image for Gender Classification. American International Journal of
Research in Science, Technology, Engineering and Mathematics 1(1),
2013,46-50.

[3]

V. J. Pillay, Textbook of Forensic Medicine and Toxicology (15th Ed.

Hyderabad: Paras Medical publishers, 2009),53-94.

[4]

Nithin MD, Manjunatha B, Preethi DS, Balarj BM "Gender

differentiation by finger ridge count among South India Population" J
FORENSIC LEG MED. 2011 FEB; 18(2):79-81. EPUB 2011 JAN 23

[5]

Ahmed Badawi, Mohamed Mahfouz, Rimon Tadross, Richard Jantz,

"Fingerprint-Based Gender Classification." Biomedical Engineering
department,University of Tennessee Knoxville.

[6]

Murlidhar Reddy Sangam,Karumanchi Krupadanam,Kol1a Anasuya "A

Study of Finger Prints: Bilateral Asymmetry and Sex Difference in the
Region of Andhra Pradesh"' Journal of Clinical and Diagnostic Research.
2011 June,Vol-5(3): 597-600

[7]

Jen feng wang, et ai, "Gender Determination using Fingertip Features"'.

Internet Journal of Medical Update 2008 Jul-Dec;3(2):22-8.

[8]

Ramanjit Kaur, Rakesh K. Garg "Detennination Of Gender Differences

From Fingerprint Ridge Density In Two Northern Indian Populations"
Problems of Forensic Sciences 2011, vol. LXXXV, 5-10 by the
Institute of Forensic Research

[9]

Acree, M. "Is there a gender difference in fingerprint ridge density?"

Forensic Science International 1999 May; 102 (1): 35-44.

[10] Ravi Wadhwa, Maninder Kaur, Dr. K.V.P. Singh "Age and Gender
Determination from Finger Prints using RVA and dct Coetlicients"' Issue
8 (August. 2013)