Evaluation the Impact of Photo Overlap Ratio on Point Cloud Density and Visual Accuracy of 3D Models in Structure-from-Motion Photogrammetry

Abstract

The accuracy of Structure from Motion (SfM) algorithm products is influenced by several factors, including the Photo overlap ratio, the image scale, Number of control points, distribution of these points, resolution of captured images, and other factors. This study examines the impact of Photo overlap ratio in consecutive images on point cloud density and visual accuracy in 3D models using the Structure from Motion (SfM) algorithm, while minimizing the influence of other factors. The results showed that an overlap ratio of 86% provides an optimal balance between accuracy and time efficiency, achieving a point cloud density of 1,239,627 points with a processing time of 4 minutes and 4 seconds. This accuracy was achieved with a lower time cost compared to the 93% overlap ratio, which increased processing time by 30% without a significant improvement in density or accuracy. It was also observed that overlap ratios lower than 86% lead to a decline in image triangulation quality and an increase in distortions. Therefore, an overlap ratio of 86% is considered optimal for photogrammetric applications that require both accuracy and speed. The study recommends future research to examine the impact of control point distribution to further improve results.