Comparative analysis of metaheuristic algorithms for modeling 3D point clouds
Advances in Space Research, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.asr.2026.08.010
- Dergi Adı: Advances in Space Research
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Artic & Antarctic Regions, Compendex, INSPEC, MEDLINE, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO)
- Anahtar Kelimeler: 3D point clouds, Data classification, Geoid determination, Metaheuristic algorithm, Optimization
- Ondokuz Mayıs Üniversitesi Adresli: Evet
Özet
Modeling 3D point clouds involves representing the surface of an environment or object with an appropriate model. The geoid is a reference surface that best represents the Earth’s surface, expressing the curvature of the surface relative to sea level and the gravity potential surface at various points. Determining the geoid has always been a subject of scientific inquiry. Numerous methods for determining the geoid have been developed, including the Global Navigation Satellite System (GNSS) Levelling. In GNSS Levelling performed using a 3D point cloud, the data used must be cleared of outlier measurements. One promising approach to improving measurement classification is the use of metaheuristic algorithms (MA), which offer flexibility and the potential to improve classification accuracy while reducing RMSE.In this study, data classification was performed using MA, which have not been used previously in geoid determination in geomatics engineering. The primary objective is to overcome the limitations of existing methods in geoid surface determination and data classification and to provide more effective solutions. In the process of geoid surface determination, the dataset contains not only points that belong to the surface but also outlier points with errors that are not representative of the surface. To obtain an accurate surface model, it is necessary to classify the points as consistent or inconsistent and to eliminate the data affected by outliers. Four MA were selected for GNSS Levelling- geoid data classification; Shuffled Frog Leaping Algorithm (SFLA), Firefly Algorithm (FA), Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO). The results are compared with the GNSS Levelling geoid determination method. Geoid determination was performed using the GNSS Levelling method. During this process, data classification was conducted through outlier measurement analysis with Conventional method on Least Square Method (LSM). The performances of the methods were evaluated by Root Mean Square Error (RMSE) and the number of common points they obtained with each other. For this study, a 3D point cloud dataset containing 8750 points was collected via a drone flight in Aşağıbeşpınar village, Sungurlu district, Çorum Province. An analysis of the results revealed that the FA, with 1,628 common points and an RMSE of 0.1323 m, produced outcomes most comparable to the LSM. With an agreement rate of 62.02%, FA showed the highest level of agreement with the LSM optimization results while also achieving the best RMSE performance, followed by the SFLA algorithm with an agreement rate of 50.36%. The agreement rates of the GWO and PSO algorithms were calculated as 48.61% and 41.64%, respectively. Based on these findings, PSO showed the lowest level of agreement among those tested. Overall, the results suggest that FA, SFLA, and GWO can compete with conventional methods and produce reliable outcomes.