p–ISSN: 2723 - 6609 e-ISSN: 2745-5254

Vol. 5, No. 8 August 2024 http://jist.publikasiindonesia.id/

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3133

Automatic Reading System of Scale Values on Webcam-

Based Measuring Vessel

Budi Yasri

, Gianto

, Rossi Aulia Rachmawati

Akademi Metrologi dan Instrumentasi, Indonesia

Email:

[email protected]

*Correspondence

ABSTRACT

Keywords: reading error,

automatic reading,

measuring vessel, image

processing, webcam

camera.

The fuel measurement pump is an installation used to

measure the quantity of liquid fuel (BBM) to be sold or

delivered to consumers. In the transaction of buying and

selling fuel at gas stations, the fuel measurement pump plays

a crucial role. Therefore, this measuring device needs to be

calibrated or re-verified, and supervision is carried out using

a standard measuring vessel with capacities of 10 L or 20 L.

Errors in reading the measuring vessel can occur due to

differences in individual visual acuity. To address this issue,

a prototype for automatic reading of measuring vessels is

created using a webcam camera. This prototype utilizes the

Raspberry Pi 3 B+ as an image processor and the Logitech

C270 webcam as an image capturer. The image processing

process involves segmentation steps, transforming the image

into a grayscale and then into a binary image. This research

involves two different types of fluids in the volume

measurement of the measuring vessel, namely distilled water

(equates) and petalite. The measurement results with the

prototype show an average accuracy level of around 99,23%,

an average precision level of about 99,37%, and an average

error rate of about 0,77%. For the equated fluid, the

accuracy, precision, and error levels are 99,48%; 99.76%;

and 0,52%, respectively. Meanwhile, for the petalite fluid,

the accuracy, precision, and error levels are 98,98%;

98,80%; and 1,02%, respectively.

Introduction

Fuel demand in Indonesia continues to increase in line with the growth of motor

vehicles and the spread of petrol stations (Putri, Novita, & Maskar, 2022). In fuel

transactions, measurement accuracy is an important factor. For this reason, fuel gauge

pump measuring instruments must be marked and re-marked as well as periodic

supervision (Sinaga, 2017). This activity is usually carried out by volumetric method

using a standard measuring vessel (BUS) as the standard. A measuring vessel is a volume-

measuring instrument that serves as a reference in testing other volume-measuring

instruments (Buana, 2018).

Budi Yasri, Gianto, Rossi Aulia Rachmawati

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3134

However, differences in the ability to read the BUS scale by the individual eye can

result in reading errors (Pertiwi & Setyawan, 2020). Therefore, an automatic reading

prototype using the camera sensor is required. Previous research has tried using ultrasonic

sensors and capacitive sensors, but it still has limitations (Adriansyah, GM, & Yuliza,

2014).

The study used camera sensors to capture BUS-scale images, which were then

processed automatically (Irmayanti & Andini, 2019). This enables accurate automatic

readings and generates evidence in the form of images for surveillance activities. The test

results show that this prototype is effective in improving the accuracy of BUS

measurements at petrol stations (Laoli, 2018). Thus, the use of camera sensor technology

can be a solution to improve measurement accuracy in fuel transactions at petrol stations

and facilitate supervision by metrology supervisors.

Figure 1. System block diagram

Research Methods

In the prototype of the reading of the scale value on the measuring vessel, a method

was carried out by simulating synthetic images. The synthetic imagery is made to

resemble the image of a measuring vessel captured by a webcam. This image is made to

resemble the colour of a petalite liquid, which is green.

Figure 2. Synthetic imagery

Figure 3. Image of the acquisition using a webcam camera

Automatic Reading System of Scale Values on Webcam-Based Measuring Vessel

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3135

The image processing techniques used are as follows:

1. Pre-processing, preparing the correct camera position and determining the appropriate

ROI (Region of Interest) to be detected by the camera.

2. The segmentation process, changing the RGB image to a grayscale image which is

then changed again to a binary image (black and white).

3. Post-processing, image analysis by detecting black pixels using the principle of liner

interpolation then calculating what volume the meniscus point is at with equation (1)

as follows.

Results and Discussion

Simulation Results

In this study, a simulation was carried out using synthetic images as shown in Figure

2 which aims to determine the reliability of the program that has been made. This

synthetic image is made to resemble the original image taken using a webcam camera. In

the synthetic image that has been made, the ROI image contains at least a liquid meniscus

object in the glass of the measuring vessel and scale lines on the glass wall, which is -50

mL to 50 mL (Zahara, 2019).

After that, the synthetic image that has been created is tested using the programming

in the prototype to obtain the results of the volume reading on the measuring vessel. In

synthetic images, the meniscus line is at a volume of 25 mL. Meanwhile, in prototype

programming, it reads 25 mL (Alfonsius et al., 2023). This indicates that the prototype

can run well because it has successfully read artificial images. The reading results have a

difference of 0 mL.

This image processing uses several stages of image processing. First, the

segmentation stage by changing the RGB colour image to grayscale or grayscale.

Figure 4 Segmentation of RGB images to grayscale images

After that, for the image to be at a certain threshold, the adaptive threshold method

is carried out. This method is performed to convert a grayscale image into a binary (black-

and-white) image taking into account the pixel intensity in about each pixel. This helps

in overcoming uneven lighting variations across the imagery (Abdi, 2020). In this process,

the grayscale image will be changed to black or white only, where black will have a binary

value of 0 while white will have a value of 1. This is shown in the meniscus producing

Budi Yasri, Gianto, Rossi Aulia Rachmawati

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3136

the colour black. In programming, the detected black colour is considered a black pixel.

The detected black pixels are marked with a red dot that has x coordinates of 290.

Figure 5. Results of segmentation of grayscale images to binary images

Next, linear interpolation is carried out to determine the volume read on the glass

glass of a standard measuring vessel. This interpolation is a method for estimating the

value between 2 previously known points on a straight line of black pixels of scale. The

reference used to determine the value of the detected meniscus black pixel is the black

pixel on the artificial scale.

Figure 6. Reading results

It is known that the value of the px coordinate is 193 pixels. In addition, the start

and end of the pixels are 193 and 211 pixels, and then the start and end of the mL values

are 10 mL and 5 mL. Thus, the value of the meniscus point in the volume is obtained as

follows.

Volume = 10 mL + (5 -10) mL x (193 - 193) Px / (211 - 193) Px

Volume = 10 mL - 5 mL x (0) / (18)

Volume = 10 mL - 0 mL

Volume = 10 mL

The simulations that have been carried out using synthetic images, show that the

image processing carried out has succeeded in reading the volume of liquid on the

measuring vessel. The results of the prototype reading showed the same results as the

Automatic Reading System of Scale Values on Webcam-Based Measuring Vessel

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3137

synthetic images that had been made. In addition, synthetic images have also been made

at other meniscus points and produced an error range of 0 mL at 11 meniscus points,

namely 25 mL to -25 mL with a difference of 5 mL at each point.

Experiment Results

Synthetic image processing has been successfully carried out. Furthermore, the

researcher conducted a direct test with real images using 2 different liquids, namely

equates and petalite. Viewed in Figure 7, a real image with liquid equates reads 10 mL.

That way, the results read by the prototype are not much different from the manual

reading, which is 10 mL.

(a) (b) (c) (d)

Figure 7

Aquades (a) RGB imagery (b) Grayscale imagery (c) Binary imagery (d) Imagery

(a) (b) (c) (d)

Figure 8

Pertalite (a) RGB image (b) Grayscale image (c) Binary image (d) Image processing

results

Viewed in Figure 8, a real image with petalite liquid reads 10 mL. That way, the

results read by the prototype are not much different from the manual reading, which is 10

mL.

Conducting experiments using 2 different liquids, will prove the reliability of the

prototype in taking volume readings on different liquid colours. In addition, petalite is a

type of fuel so it has different properties and densities compared to equates. In the

calibration of the measuring vessel, equate liquid is used. That way, in addition to being

able to be used for surveillance and re-stamping purposes, this prototype can also be used

for calibration purposes.

Conclusion

In this study, a prototype of an automatic reading system for scale values on a

measuring vessel using a webcam camera has been made, where this prototype can

automatically read the volume of the measuring vessel on equates and petalite liquids

through the integration of prototype-supporting hardware and software to be able to detect

meniscus points through image processing with OpenCV Python.

Method validation for the prototype has been carried out by comparing the readings

on the prototype with manual readings or references using a nonius scale of a measuring

Budi Yasri, Gianto, Rossi Aulia Rachmawati

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3138

vessel with an accuracy of 0.5 mL and an artificial line with an accuracy of 0.05 mL.

Based on the results of the method validation, the prototype reading has approached the

manual reading.

The measurement of the volume of the measuring vessel with the automatic reading

prototype has an average accuracy rate of 99.23%; the average precision is 99.28%; and

the average error is 0.77%. In equates liquid, the level of accuracy, precision, and error is

99.48%; 99,76%; and 0.52%. Meanwhile, in petalite liquid, the level of accuracy,

precision, and error is 98.98%; 98,80%; and 1.02%. Not only used for surveillance

activities but the prototype can also be used for calibration of measuring vessels because

the prototype can take readings with adequate liquid.

Automatic Reading System of Scale Values on Webcam-Based Measuring Vessel

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3139

Bibliography

Abdi, Paki Abror. (2020). Pengaruh Kualitas Pelayanan Terhadap Kepuasan Konsumen

Pada SPBU Zacky Muhammad Pekanbaru. Universitas Islam Riau.

Adriansyah, Andi, GM, Mirzanu Rizki, & Yuliza, Yuliza. (2014). Rancangbangun dan

Analisa Cctv Online Berbasis Raspberry Pi. SINERGI, 18(2), 105–110.

Alfonsius, Eric, Johannes, Albani Boutje, Mantiri, Roma Nouke Frets, Manahampi,

Romei, Hihola, Maharani, & Hadiwidjaja, Angelica Christy. (2023). Sistem

Informasi Geografis Letak Persebaran Tempat Pengisian Bahan Bakar Kendaraan

Tingkat Retailer Menggunakan Google API. Information System Journal, 6(02), 76–

85.

Buana, I. Komang Setia. (2018). Aplikasi untuk pengoprasian komputer dengan

mendeteksi gerakan menggunakan opencv python.

Irmayanti, Tiara, & Andini, Khairunnisa Desi. (2019). Prototipe kontrol level otomatis

berbasis sensor ultrasonik dan variabel speed control pompa pada pengisian bejana

ukur standar. Akademi Metrologi dan Instrumentasi.

Laoli, Desimeri. (2018). Penerapan Algoritma Hill Cipher Dan Least Significant Bit

(LSB) Untuk Pengamanan Pesan Pada Citra Digital. Jurnal Ilmu Komputer Dan

Sistem Informasi (JIKOMSI), 1(2), 30–40.

Pertiwi, Mas Ninda, & Setyawan, Galih. (2020). Bejana Ukur Standar Dengan Tampilan

Digital Menggunakan Sensor Kapasitif Berbasis Arduino Nano. Simetris: Jurnal

Teknik Mesin, Elektro Dan Ilmu Komputer, 11(2), 437–447.

Putri, Almira Devita, Novita, Dian, & Maskar, Sugama. (2022). Pengenalan Wawasan

Bisnis Di Era Digital Bagi Siswa/I Smk Yadika Bandarlampung. Journal of Social

Sciences and Technology for Community Service (JSSTCS), 3(2), 213–217.

Sinaga, Anita Sindar R. M. (2017). Implementasi Teknik Threshoding Pada Segmentasi

Citra Digital. Jurnal Mantik Penusa, 1(2).

Zahara, Alfan Nawaziru. (2019). Penentuan kadar emas menggunakan metode

interpolasi Lagrange. Universitas Islam Negeri Maulana Malik Ibrahim.