Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads
Land transportation accidents are a severe problem in Indonesia. Factors that cause land transportation accidents include driver negligence, not road-worthy vehicles, and damaged road conditions. In 2018, according to data from the Indonesian Central Statistics Agency, the number of accidents in Ind...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-12-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123023005649 |
| _version_ | 1827578685392158720 |
|---|---|
| author | Fitri Utaminingrum Ainandafiq Muhammad Alqadri I Komang Somawirata Corina Karim Anindita Septiarini Chih-Yang Lin Timothy K. Shih |
| author_facet | Fitri Utaminingrum Ainandafiq Muhammad Alqadri I Komang Somawirata Corina Karim Anindita Septiarini Chih-Yang Lin Timothy K. Shih |
| author_sort | Fitri Utaminingrum |
| collection | DOAJ |
| description | Land transportation accidents are a severe problem in Indonesia. Factors that cause land transportation accidents include driver negligence, not road-worthy vehicles, and damaged road conditions. In 2018, according to data from the Indonesian Central Statistics Agency, the number of accidents in Indonesia due to damaged roads was around 36,89%. One of the types of road damage is potholes. Potholes have the potential to trigger road accidents, especially for motorcyclists. By looking at current technological developments, the Unmanned Ground Vehicle (UGV) is a transportation technology that does not have a crew that can help detect obstacles on the highway, such as potholes. The potholed road itself has a different texture from normal roads. The pothole texture value can be represented using Gray-Level Cooccurrence Matrix (GLCM) as a feature extraction algorithm. GLCM has several features and combinations, namely the distance and angle features of GLCM. Too many features are difficult to implement in artificial intelligence because it requires a long computation time. The system needs to produce a fast computation time in detecting the pothole so that it can be implemented in real-time. A Genetic Algorithm is applied to perform feature selection. Furthermore, the classification method is used Extreme Learning Machine (ELM). The GLCM features used are 128 features which will be tested three times with the selection results obtained as a combination of 57 features, 20 features, and 12 features. Based on the best variety of features in the video test, the accuracy is a combination of features 57 of 88,65% and a computation time of 0,115 s. The best overall conclusion is the combination of feature selection 20 and 12 because it has an accuracy of 87,36% and 86,48%, which is not much different from feature selection 57. However, it has a much faster computation time of 0,069 s and 0,062 s. |
| first_indexed | 2024-03-08T21:49:59Z |
| format | Article |
| id | doaj.art-60c2cb27f830483a8d9d49aa4ca96dc4 |
| institution | Directory Open Access Journal |
| issn | 2590-1230 |
| language | English |
| last_indexed | 2024-03-08T21:49:59Z |
| publishDate | 2023-12-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj.art-60c2cb27f830483a8d9d49aa4ca96dc42023-12-20T07:35:42ZengElsevierResults in Engineering2590-12302023-12-0120101437Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roadsFitri Utaminingrum0Ainandafiq Muhammad Alqadri1I Komang Somawirata2Corina Karim3Anindita Septiarini4Chih-Yang Lin5Timothy K. Shih6Computer Vision Research Group, Faculty of Computer Science, Brawijaya University, Malang, 65145, Indonesia; Corresponding author.Computer Vision Research Group, Faculty of Computer Science, Brawijaya University, Malang, 65145, IndonesiaDepartment of Electrical Engineering, National Institute of Technology, Malang, IndonesiaMathematics Department, Brawijaya University, Jl. Veteran Malang, East Java, IndonesiaDepartment of Informatics, Faculty of Engineering, Mulawarman University, Samarinda, 75119, IndonesiaComputer Vision and Interactive Technology, Department of Electrical Engineering, Yuan-Ze, University, Taoyuan City, TaiwanDepartment of Computer Science and Information Engineering, National Central University, Taoyuan City, 32001, TaiwanLand transportation accidents are a severe problem in Indonesia. Factors that cause land transportation accidents include driver negligence, not road-worthy vehicles, and damaged road conditions. In 2018, according to data from the Indonesian Central Statistics Agency, the number of accidents in Indonesia due to damaged roads was around 36,89%. One of the types of road damage is potholes. Potholes have the potential to trigger road accidents, especially for motorcyclists. By looking at current technological developments, the Unmanned Ground Vehicle (UGV) is a transportation technology that does not have a crew that can help detect obstacles on the highway, such as potholes. The potholed road itself has a different texture from normal roads. The pothole texture value can be represented using Gray-Level Cooccurrence Matrix (GLCM) as a feature extraction algorithm. GLCM has several features and combinations, namely the distance and angle features of GLCM. Too many features are difficult to implement in artificial intelligence because it requires a long computation time. The system needs to produce a fast computation time in detecting the pothole so that it can be implemented in real-time. A Genetic Algorithm is applied to perform feature selection. Furthermore, the classification method is used Extreme Learning Machine (ELM). The GLCM features used are 128 features which will be tested three times with the selection results obtained as a combination of 57 features, 20 features, and 12 features. Based on the best variety of features in the video test, the accuracy is a combination of features 57 of 88,65% and a computation time of 0,115 s. The best overall conclusion is the combination of feature selection 20 and 12 because it has an accuracy of 87,36% and 86,48%, which is not much different from feature selection 57. However, it has a much faster computation time of 0,069 s and 0,062 s.http://www.sciencedirect.com/science/article/pii/S2590123023005649AccidentPotholesGLCMGAELM |
| spellingShingle | Fitri Utaminingrum Ainandafiq Muhammad Alqadri I Komang Somawirata Corina Karim Anindita Septiarini Chih-Yang Lin Timothy K. Shih Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads Results in Engineering Accident Potholes GLCM GA ELM |
| title | Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads |
| title_full | Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads |
| title_fullStr | Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads |
| title_full_unstemmed | Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads |
| title_short | Feature selection of gray-level Cooccurrence matrix using genetic algorithm with Extreme learning machine classification for early detection of Pole roads |
| title_sort | feature selection of gray level cooccurrence matrix using genetic algorithm with extreme learning machine classification for early detection of pole roads |
| topic | Accident Potholes GLCM GA ELM |
| url | http://www.sciencedirect.com/science/article/pii/S2590123023005649 |
| work_keys_str_mv | AT fitriutaminingrum featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT ainandafiqmuhammadalqadri featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT ikomangsomawirata featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT corinakarim featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT aninditaseptiarini featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT chihyanglin featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads AT timothykshih featureselectionofgraylevelcooccurrencematrixusinggeneticalgorithmwithextremelearningmachineclassificationforearlydetectionofpoleroads |