Utilization of Open-Source Photogrammetry Software with Structure from Motion Algorithm for Three- Dimensional Modeling

Reconstructing a three-dimensional model of a real-world object requires a quality point cloud to produce an accurate three-dimensional model. Point clouds can be produced by image-based and distance-based methods. The image-based method utilises the light reflected by the object, which is then processed using the principle of photogrammetry. This research is intended to test the image-based method with the help of the structure from motion approach to generate the quality of point clouds using open-source software including Visual SfM, MicMac, Meshroom and with the Agisoft Metashape Pro commercial software. The images are taken by a Xiaomi Yi Action camera with 2.68 mm focal length. The performance comparisons will consider Volume Density, Surface Variation and Point discrepancy. We utilize BLK 360 terrestrial laser scanner to collect reference point cloud. The results show that all software provide convincing result on each parameters compared.