Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry
Intensive fruit or vegetable sorting is a common mission in productive regions. In order to meet the market standards, it is produced according to quality levels that depend on maturity, weight, size, density, skin defects and others. The aim of this study is to automate the sort of oranges based on...
Published in: | International Journal of Geoinformatics |
---|---|
Main Author: | |
Format: | Article |
Language: | English |
Published: |
Association for Geoinformation Technology
2022
|
Online Access: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85135169581&doi=10.52939%2fijg.v18i4.2251&partnerID=40&md5=44ed8f08964556cd810432fc16f03142 |
id |
2-s2.0-85135169581 |
---|---|
spelling |
2-s2.0-85135169581 Osman N.S.; Tahar K.N.; Almhafdy A.A. Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry 2022 International Journal of Geoinformatics 18 4 10.52939/ijg.v18i4.2251 https://www.scopus.com/inward/record.uri?eid=2-s2.0-85135169581&doi=10.52939%2fijg.v18i4.2251&partnerID=40&md5=44ed8f08964556cd810432fc16f03142 Intensive fruit or vegetable sorting is a common mission in productive regions. In order to meet the market standards, it is produced according to quality levels that depend on maturity, weight, size, density, skin defects and others. The aim of this study is to automate the sort of oranges based on colour and size to specific grades. The experiment on the oranges colour grading using K-Means and the Isodata technique has been used in this study. The methodology consists of four phases which are data acquisition, data pre-processing, data segmentation and data analysis. Moreover, the results of the RGB value and measurement size of the oranges are based on the output that has been achieved. The RGB value is to determine the maturity of the oranges based on the RGB value. In addition, the measurement size of the oranges was computed using the 3DF Zephyr that generated the 3D images. The manual measurement and direct measurement using photogrammetry software will be able to compare the different sizes of the oranges. Based on the results obtained, the best type of oranges that represents Grade A by size and colour output is Grapefruit Australia. The reason is that the size measurement using a comparison between manual and the 3D builder tools shows the best outcome that, is a 0.1 cm difference, while the manual and 3D Zephyr difference is 4.28-cm. Consistent RGB percentage for both classifications show good ripe. The red colour percentage is 37.6%, the green colour is 37.5%, and the blue colour is 16.2%. Besides that, all the oranges are also supported with depth accuracy and a ground pixel size under the 3D images technique that is under the tolerance of one-pixel disparity. In conclusion, the grading of orange is determined using the size measurement from 3D images and colour classification using the unsupervised classification of K-Means or Isodata Classification. © Geoinformatics International. Association for Geoinformation Technology 16866576 English Article All Open Access; Bronze Open Access |
author |
Osman N.S.; Tahar K.N.; Almhafdy A.A. |
spellingShingle |
Osman N.S.; Tahar K.N.; Almhafdy A.A. Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
author_facet |
Osman N.S.; Tahar K.N.; Almhafdy A.A. |
author_sort |
Osman N.S.; Tahar K.N.; Almhafdy A.A. |
title |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
title_short |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
title_full |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
title_fullStr |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
title_full_unstemmed |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
title_sort |
Machine Learning Approach in Orange Fruit Grading Using Close Range Photogrammetry |
publishDate |
2022 |
container_title |
International Journal of Geoinformatics |
container_volume |
18 |
container_issue |
4 |
doi_str_mv |
10.52939/ijg.v18i4.2251 |
url |
https://www.scopus.com/inward/record.uri?eid=2-s2.0-85135169581&doi=10.52939%2fijg.v18i4.2251&partnerID=40&md5=44ed8f08964556cd810432fc16f03142 |
description |
Intensive fruit or vegetable sorting is a common mission in productive regions. In order to meet the market standards, it is produced according to quality levels that depend on maturity, weight, size, density, skin defects and others. The aim of this study is to automate the sort of oranges based on colour and size to specific grades. The experiment on the oranges colour grading using K-Means and the Isodata technique has been used in this study. The methodology consists of four phases which are data acquisition, data pre-processing, data segmentation and data analysis. Moreover, the results of the RGB value and measurement size of the oranges are based on the output that has been achieved. The RGB value is to determine the maturity of the oranges based on the RGB value. In addition, the measurement size of the oranges was computed using the 3DF Zephyr that generated the 3D images. The manual measurement and direct measurement using photogrammetry software will be able to compare the different sizes of the oranges. Based on the results obtained, the best type of oranges that represents Grade A by size and colour output is Grapefruit Australia. The reason is that the size measurement using a comparison between manual and the 3D builder tools shows the best outcome that, is a 0.1 cm difference, while the manual and 3D Zephyr difference is 4.28-cm. Consistent RGB percentage for both classifications show good ripe. The red colour percentage is 37.6%, the green colour is 37.5%, and the blue colour is 16.2%. Besides that, all the oranges are also supported with depth accuracy and a ground pixel size under the 3D images technique that is under the tolerance of one-pixel disparity. In conclusion, the grading of orange is determined using the size measurement from 3D images and colour classification using the unsupervised classification of K-Means or Isodata Classification. © Geoinformatics International. |
publisher |
Association for Geoinformation Technology |
issn |
16866576 |
language |
English |
format |
Article |
accesstype |
All Open Access; Bronze Open Access |
record_format |
scopus |
collection |
Scopus |
_version_ |
1809677891254878208 |