3D Models Allow the Recovery of Information from Video Transects: A Case Study in Coral Reefs

Abstract

Image data acquisition is crucial for understanding long-term ecological processes. The video transect technique, developed in 1996 in Australia, has the main advantage of continuously visual recording the reef, allowing for the recovery of information that was previously unrecognized as relevant. With technological advances, the recovery of this information has reached a new level. Using photogrammetric techniques, it is possible to three-dimensionally reconstruct the filmed surface of reefs, enabling the collection of fundamental data on the structural complexity and population structure of corals and other benthic species. Structural complexity is the number of irregularities and indentations in the reef, indicating the amount of shelter it provides to marine fauna. Population structure describes the age distribution of species and is positively correlated with structural complexity. In this paper, we present the application of a 3D photogrammetry method to measure structural complexity in coral reefs using video transect images. The video transects used are 20 meters long by 48 centimeters wide, with an average duration of two to three minutes. To construct the models, 350 deinterlaced images were extracted from each video, ensuring a high overlap rate. The models were generated using Metashape 2.0.0 (Agisoft). The images were automatically aligned, followed by the generation of a mesh of overlapping points, which was then textured. A mosaic model was then produced. Spatial scale was calibrated using ten one-centimeter markings arranged along a 20-meter measuring tape. Structural complexity was estimated by the ratio between the number of points and vertices and the reconstructed area of the transect. To ensure comparability between models, it is essential to standardize the parameters used during processing to avoid bias. Changes in camera lens, resolution, and image quality can also affect the results. Despite these limitations, the technique offers a valuable opportunity to revisit the past of ecosystems recorded on video, allowing retrospective analyses with a high level of detail.