Abstract
The semi-automatic and automatic extraction of land features such as buildings, trees, and roads using aerial laser scan data is crucial in land use change studies and urban management. This research introduces the "BTR" extractor, a novel software package designed to enhance classification accuracy of phenomena identified in the super points obtained from aerial laser scanners. Our method focuses on:-Comparing classification methods using airborne laser scanning data.-Implementing supervised algorithms for high-accuracy classification.-Evaluating the performance against existing software like TerraSolid. The user-friendly interface allows data entry, training data collection, and selection of classification methods. We employed five methods (Bayesian algorithms, support vector machine, K-nearest neighbor, C-Tree, and discriminant analysis) to classify land features. Comparative results show the BTR extractor outperforms TerraSolid, particularly in supervised classification, demonstrating high accuracy and reliable implementation in the studied area. Our findings advocate for the use of supervised algorithms in classifying cloud data for enhanced accuracy and efficiency in remote sensing applications.