Application of existing algorithms for classification and clustering of laser reflection points (k-Means, DBSCAN, SVM) to solve territorial planning problems

Abstract

Objective. The aim of the study is to generalize the experience of using and reveal the features of methods for classifying and clustering images obtained by laser scanning.Method. The study is based on the use of algorithms for recognition, clustering and classification of data obtained using laser scanning.Result. A brief review of existing algorithms used for clustering images obtained by laser scanning has been carried out. The possibility of using sequentially the K-means and DBSCAN algorithms for clustering in relation to objects of various shapes is shown. The possibilities of using algorithms for the classification and clustering of laser reflection points (k-Means, DBSCAN, SVM) in the framework of solving territorial planning problems are analyzed.Conclusion. The use of the presented algorithms makes it possible to form data arrays that, when superimposed on each other subsequently and further processed, make it possible to obtain even more accurate representations of objects and territories in territorial planning documents, and in the future, based on the use of machine learning methods and processing of data matrices available in memory, get more accurate information about objects on the ground.

Keywords

• territorial planning
• k-means
• dbscan clustering
• classification
• point cloud
• laser scanning