Body Mass Index Measurement System using Image Processing
Abdullah Sani and Elsa Agatya Marlin
Department of electrical Engineering, Politeknik Negeri Batam, Batam, Indonesia
Keywords: Image Processing, Broca Formula, Body Mass Index.
Abstract: In general, to know the ideal body weight, one must first weigh the weight and measure its height manually.
After that do the calculations with certain mathematical formulas so that a conclusion about the condition is
ideal body weight. However, it is less efficient especially if applied in bulk. Therefore, the author will create
a system with the Image Processing application in real-time using the webcam by specifying the height and
width of the body at a specific distance and scale that applies the calculation of the value of the pixel values
where the pixel value is as a reference as a variable function to obtain the person's height and weight estimate.
Then after gaining height and weight than the calculation of Body Mass Index with Broca formula to be useful
as an identification system in effectively measuring human body mass index. This system can be used as a
system that can require humans to more efficient time and reduce human error. The output generated from
this system is the high value, weight, and BMI values of the person's body whether the condition is ideal or
not.
1 INTRODUCTION
This In the current era of developments, various
human-created technologies are required to be more
effective, fast, and easy in assisting human work in
the areas of health, information, communication, and
other areas. The more technology used is the more
practical human work. The Discovery and
development of new technologies are expected to
bring a positive impact on human life in all things
more efficiently (Triyandi, 2014).
The identification system of a person does not
escape the technological developments that occurred
in the present era. Such an identification system can
be a measurement of the human body by utilizing
high parameters and weight gain (Wicaksana 2016).
By utilizing these parameters, one can identify and
know the condition of the body is ideal or not. In
general, to know the ideal body weight, one must first
perform the measurements manually. After that do
the calculations with certain mathematical formulas
so that a conclusion about the condition is ideal body
weight. However, there are still many people who
rarely control their weight regularly, while it is very
important to know whether the condition of the body
is good or not in the weight and height, especially if
applied in bulk because it will require a lot of
weighing and measuring tools as well as long enough
time for its use (Efendi, 2017).
It is therefore created a system in real-time using
the webcam by specifying the height and width of the
body at a specified distance and scale that applies the
calculation of the pixel value in which the pixel value
is referenced as a variable function to obtain the
person's height and weight estimate.
Then after gaining height and weight, the
calculation of Body Mass Index with Broca formula
to be useful as an identification system in measuring
the human body mass index.
2 BASIC UNDERSTANDING
2.1 Image Processing
Image processing or image processing is a method to
better improve the image quality by utilizing several
methods of image processing itself, such as the setting
on the color or image size. Image processing is widely
utilized to detect an object, where input from the
image processing can be a photo or motion picture in
real-time, while the output can be a parameter related
to imagery such as the pixel value in the image
detected according to its needs (Putri, 2016).
Sani, A. and Marlin, E.
Body Mass Index Measurement System using Image Processing.
DOI: 10.5220/0010352000810084
In Proceedings of the 3rd International Conference on Applied Engineering (ICAE 2020), pages 81-84
ISBN: 978-989-758-520-3
c
81
Research is conducted by implementing the
image processing system in the calculation process,
where when a person's body is faced with the camera,
the camera will detect the image (tracking) and then
determine the point of the coordinate until finally
obtained the high value and width of the pixel used as
a reference in the calculation.
2.1.1 RGB Image
The RGB image or color image is a digital image
composed of 3 main colors, namely red, green, and
blue. Meanwhile, in an image, other colors are the
result of the combination of the three main colors
(Makalalang, 2012). RGB color concepts can be
shown in Figure 1 RGB Color Concepts.
2.1.2 Blob Counter
Blob Counter is a method of image processing used
to calculate, filter, and extract objects contained in an
image by mapping the object to obtain the dimensions
or size of the desired object. The principle is to nod
all the pixels that have a value less than the same as
the background pixel value as the background of the
object. Thus the pixel that has the value above the
background pixel value will be considered as the
object. The mapping is then done to distinguish
between the background and object (Uma, 2010).
2.1.3 Broca Formula
The BROCA formula is a formula used to determine the
ideal body weight by simply utilizing the height
measurement parameters (Hutabarat, 2016). In this
study, using a formula of Broca that is utilized to get the
value of a person's ideal weight and also is one as a point
of reference to get the value of weight prediction from
the detected body. Here is the Broca formula
Ideal BW
Kg
=
Body Height
cm
-100
-
Body Height
cm
-100
×10%
(1)
Figure 1: RGB color concepts (Makalalang, 2012).
2.1.4 Body Mass Index
Body Mass Index is an indicator to determine the
bodyweight of a person whether it has a body that is
too thin, ideal, or too fat and is a good measurement
way to assess the risk of illness that can occur due to
an unideal weight, by utilizing the height parameters
in meters and weight in units Kg (Saffana, 2018;
Andre, 2016; Heriansyah, 2017). The formula of the
body mass index calculation can be seen in Equation
(2).
BMI=
Body Weight
Kg
Bod
y
Hei
g
ht
m
×Bod
y
Hei
g
ht
m
(2)
From the result of the calculation, it is obtained by
classifying the percentage of whether a person's
weight is too thin, ideal, or too fat. The classifications
can be seen in Table 1. BMI Value Classification.
3 PROPOSED METHOD
In Figure 2, the first step is to do an acquisition of an
image to the object (human body) in a full body in a
standing position. This process is conducted with a
distance of 300 cm against the object as well as a
camera height 80 cm from the floor. In the detection
process, the camera will look for the object in real -
time, where the rectangle will enlarge or shrink
according to the detected object, and the result of the
pixel coordinate value will also change according to
the rectangle size, after the stable rectangle, then
capture the image to make it easier for someone to
know the body mass index.
When the object has been captured by the camera,
the program will then perform its approximate height
and width calculation using pixel value calculations.
Where to use the calculation formula as follows:
Body Height Prediction
cm
=
Pixel Heigh
t
-Average(Error Pixel Height)
(3)
Body Width Prediction
cm
=
Pixel Width-Average(Error Pixel Width)
(4)
Table 1: BMI value classification.
Nutritional Status BMI
Weight deficiency < 18,5
Ideal 18,5
24,9
Excess weight 25
30
Fat 30
40
Fat unhealth
> 40
ICAE 2020 - The International Conference on Applied Engineering
82
Figure 2: Flowchart system.
After the approximate value of height and width
of the body in cm, the next step is to do the calculation
of Ideal body weight with Broca formula as in
Equation 1. The next step is to determine the
approximate weight value of the body by utilizing the
height and width parameters, as well as the ideal
weight value. The calculation formula is as presented
in Equation (5). After obtaining the high prediction
and weight parameters, the Body Mass Index (BMI)
calculation with the calculation of the formula
corresponds to equation (2).
Body Weight Prediction
Kg
=
Average∆
BH-BWi
-BH-BWi
+BW Ideal(Kg)
(5)
B.H (cm): Height prediction value in cm units.
Average (Error Pixel Height): The average value
of pixel high error obtained from sample data.
Average (Error Pixel Width): The average value
of of pixel width error obtained from sample
data.
Average (∆
B.H-B.Wi
): The average value of
difference from the height prediction and boody
width obtained from sample data.
(B.H B.Wi): Delta value in height prediction and
width prediction of body detected.
4 RESULT AND DISCUSSION
4.1 Data Retrieval
At the beginning of the study, cropping on the image
filtering results by removing the background from the
image so that only the remaining objects will be
detected and then calculated. Filtering is done by
setting the RGB parameter, in this research the RGB
value is set to the value R = 222, G = 89, B = 83, then
the RGB filtering result image in the filter returned to
binary to get the intact object image so that it is easy
to detect by the BLOB counter. The result of the
BLOB counter can be shown in Figure 3.
4.2 Determination Equation of
Calculation
The determination of the equation of calculation in
this research is conducted by taking some sample
objects (human body) which are used as the
calculation of height and width estimate of the body.
Sample data is taken in the form of height, shoulder
width, and actual weight and calculation of the ideal
weight gained from the actual height and weight. The
comparison is conducted by retrieving the sample
data of the height and width pixel on the tracking
result object on the camera. The result of the
calculation will be recorded for further analysis.
The obtained data is calculated to determine the
ideal weight value of a person by processing the
existing parameters. The average value of the error at
the height of the pixel and the average value of the
error at the pixel width can be used as a reference to
get the height and width predicted value of the human
body detected by the camera. In the study, 30 human
objects were used as samples to determine the
average error value, and the value obtained for is 64.5
for the average height error and 2.25 for the average
error width of the body. Determination of calculations
to get high prediction values and width of the body
can be seen in equations 3 and 4. The following Table
2 is the result of the high and wide prediction
calculation of the body. The prediction of person's
weight prediction could be calculated using Equation
(5). The calculation results are presented in Table 3
and Table 4.
Body Mass Index Measurement System using Image Processing
83
Figure 3: Blob counter detection results.
Table 2: Results of height and width prediction calculation.
H.B
Pre.
Error
B.Wi
Pre.
Error
No
Name
Act
H.B
Act
B.Wi
(cm)
(cm)
(cm)
(cm)
(cm)
(cm)
1
Alin
165,0
165,6
0,6
40,0
36,8
3,3
2
Arya
173,0
176,6
3,6
41,0
38,8
2,3
3
Bunga
156,0
161,6
5,6
37,0
36,8
0,3
4
Cindy
162,0
158,6
3,4
40,0
35,8
4,3
5
Dita
160,0
155,6
4,4
34,0
33,8
0,3
6
Erin
156,0
149,6
6,4
38,0
36,8
1,3
7
Ilmi
160,0
157,6
2,4
43,0
40,8
2,3
8
Ipen
166,0
167,6
1,6
46,0
44,8
1,3
9
Limcol
168,0
169,6
1,6
42,0
40,8
1,3
10
Maul
174,0
180,6
6,6
43,0
38,8
4,3
Table 3: Calculation results on sample data.
Table 4: Calculation results on sample data
5 CONCLUSIONS
Based on the results and analysis that has been done
in the study can be concluded that the system can
work with an average percentage of the body's high
error rate of 4.1% and weight by 8.6 and the distance,
color, and clothing used by the measured object
greatly affect the outcome of the image.
REFERENCES
Andre, J. A., 2016. System security webcam dengan
menggunakan Microsoft visual basic. STMIK Amik,
Riau, Pekan Baru.
Efendi, T., Uljanah, I. I., Tsauri, T. A., 2017. Rancang
Bangun Sistem Pengolahan Citra Digital untuk
Menentukan Berat Badan Ideal. Jurnal Informatika
Sunan Kalijaga.
Heriansyah, T., 2017. Hubungan Indeks Massa Tubuh
dengan Jumlah Circulating Endothelial cell. Jurnal
Pendidikan dan Pengajaran.
Hutabarat, J. C. J., 2016. Perancangan Aplikasi
Menentukan Berat Badan Ideal Dengan Menggunakan
Algoritma K-MEANS CLUSTERING. Riset Komputer
(JURIKOM).
Makalalang, R. A., 2012. Perancangan Sistem Pemantau
dan Penentuan Tempat Parkir Berdasarkan Digital
Image Processing. Buku Jurnal Teknik Elektro,
Universitas Sam Ratulangi Sulawesi Utara.
Putri, A. R., 2016. Pengolahan Citra dengan Menggunakan
Webcam pada Kendaraan Bergerak di Jalan Raya.
Jurnal Ilmiah Pendidikan Informatika.
Saffana, N., 2018. Menghitung Berat Badan Ideal dengan
Rumus Broca. Indonesia.
Triyandi, D., Adler, J., 2014. Sistem Otomasi Gerbang
dengan Pengolahan Citra Membaca Nomor Plat
Kendaraan. Universitas Komputer Indonesia,
Bandung, Jawa Barat.
Uma, F., 2010. Penjejak Objek Pada Kursi Roda Cerdas
Autonomous. Makalah Seminar Tugas Akhir, Jurusan
Teknik Elektro, Universitas Dipenogoro.
Wicaksana, R. B., Hidayat, B., Aulia, S., 2016.
Classification Height and Weight of Human Footprint
Using Gabor Wavelet and KNN Method Based on
Android. e-Proceeding of Engineering, vol. 3, IEEE.
ICAE 2020 - The International Conference on Applied Engineering
84