REVIEW OF 3D RECONSTRUCTION METHODS BASED ON LASER SCANNING AND MULTI-CAMERA TRIANGULATION

Abstract

This article provides a comparative analysis of modern 3D reconstruction methods based on laser scanning (LiDAR) and multi-camera triangulation. 3D reconstruction is a vital tool in various fields such as architecture, industrial design, cultural heritage preservation, autonomous navigation, and virtual reality. Laser scanning delivers high precision and dense point clouds, making it ideal for scanning large and complex structures, yet it requires expensive equipment and longer processing times. In contrast, multi-camera triangulation, relying on the processing of images captured from several cameras or viewpoints, is effective in generating detailed 3D models of textured surfaces, although its accuracy depends significantly on lighting conditions and camera configuration.

The article examines key stages of each method, including pre-calibration, segmentation, point cloud alignment, and mesh generation. Algorithms such as Structure from Motion (SfM), Multi-View Stereo (MVS), and Iterative Closest Point (ICP) are reviewed, along with recent hybrid approaches that combine the strengths of both techniques. A comparative assessment is provided in terms of accuracy, speed, scalability, and implementation cost. Real-world applications are illustrated for each method, including digital reconstruction of architectural heritage, forensic analysis, quality control in manufacturing, and automated mapping.

The findings indicate that the choice of a 3D reconstruction technique depends on required accuracy, equipment availability, scanning environment, and object characteristics. Future directions suggest the development of hybrid systems integrating LiDAR and photogrammetry to achieve optimal accuracy and efficiency.