APPLICATION OF MACHINE LEARNING ALGORITHMS IN LIDAR POINT CLOUD CLASSIFICATION

The extraction and classification of features from three-dimensional (3D) LiDAR point clouds have become essential in photogrammetry, remote sensing, and smart city modeling. This study investigates the application of a number of supervised machine learning methods for classifying LiDAR point clouds, including Random Forest, Decision Tree, Naïve Bayes, Neural Networks, AdaBoost, Support Vector Machines (SVM), and K-Nearest Neighbours (K-NN). Experiments were conducted on two datasets with various geometric and radiometric features. Evaluation measures such F1-Score, Precision, and Recall, and computational time were analyzed. The results demonstrate the effectiveness of integrating geometric features and ensemble learning strategies in improving classification performance.