Assessing the forest's capacity to mitigate climate change and enable multi-functional management strategies for biodiversity and ecosystem service provision requires vast knowledge about the forest on a large scale. Traditional forest inventory methods can be labor-intensive and lack detailed enough measurements for large forest areas. To enable scalable forest inventories, a reliable and replicable approach for estimating key forest features with minimal manual effort and maximum impact is needed.
While terrestrial laser scanning (TLS) is considered the most precise method for generating point clouds in forestry, it can often be expensive and impractical for complex forest conditions, such as dense understory and steep slopes. In contrast, digital sensors like mass-produced cameras have become popular for data collection due to their availability and lightweight design. Additionally, videogrammetry, a method that creates 3D models from digital camera videos, has emerged as an efficient and low-cost approach for point cloud generation. However, its potential in the heterogeneous forest environment remains to be fully explored.
We propose a videogrammetry approach for reconstructing 3D point clouds in forest environments, leveraging a cost-effective and user-friendly Insta360 Pro 2 acquisition setup. This setup, incorporating six fish-eye cameras within a single camera body, allows the simultaneous capturing of six fish-eye videos and the complete coverage of a 360-degree field of view. Compared to more cumbersome alternatives, this setup simplifies and enhances data collection in various forest conditions.
We compared two videogrammetric approaches for building a point cloud in a forest environment using Insta360 Pro 2. The multi-rig approach, using videos from the six fish-eye lenses, outperformed the approach based on the stitched spherical video in reconstruction accuracy. While it required a longer computational time, the multi-rig system yielded a ground control point error of 2 mm, compared to 5 cm for the spherical point cloud, demonstrating its superiority.
