Photogrammetry

Background

Photogrammetry involves using overlapping images to construct 3D models. This has become an insightful tool in marine research, as this technique allows researchers to study underwater subjects in high detail. Indeed, it has been applied in studies ranging from monitoring coral reef health to assessing bodyt condition in free-ranging cetaceans.

In the following webpage, I demonstrate the versatility of photogrammetry by creating a 3D model of a shell, conducting a temporal analysis to examine coral growth and morphological changes over time, and perform a large-scale study involving an orthomosaic of Heron Reef to assess coral density as well as other metrics. Given the broad scope of my study, photogrammetry has served to provide ecological insights into reef health, and has potential applications in guiding conservation efforts.

Building a 3D Model of a Shell

Basic Model Details:

Processing and Results:

The main steps (from data collection to metrics extraction) of a 3D photogrammetry study involve:

Acquire imagery of the focal subject and include the necessary markers from which to create a scale bar during subsequent analysis
The images will then be imported into the Agisoft Metashape software
After undergoing an image quality check, the photos will be aligned at the lowest quality, which allows for rapid assessment of the model and adjustments prior to more computationally demanding tasks. In this step, the program estimates the position of the camera for each image, and returns a tie point cloud and a set of estimated camera positions, the latter of which are used for the 3D reconstruction process
Detect coded targets (markers) and create an associated scale bar and reference distance using known distances between each marker
Granted the low quality alignment has been completed, a high quality alignment will now be developed in order to build a very detailed 3D model (it is important to ensure >80% of the photos are aligned)
Build a point cloud (also known as a dense cloud) based on the image orientation parameters Metashape has estimated
Utilising the point cloud, build a mesh that represents the surface of the object as a series of vectors between each point, with the area in between termed faces
Build a texture to account for the colours within the photos. This will result in a more realistic 3D representation of the feature
Duplicate the chunk and use the selection tool to select parts of the model that are not necessary to include in the final version.
Close the holes in the model and calculate volume and surface area

Temporal Analysis: Growth of an individual Acropora sp. Over Time

The following dataset, provided by Dr. Renata Ferrari from AIMS, showcases structural changes in coral over time. Collected at Lizard Island in 2014 and 2015, the data offers an example of the imagery needed to analyze shifts in marine environments, particularly changes in coral morphology and structure. In this case, these images illustrate the changes in an Acropora sp. between 2014 and 2015. Kindly note, this data was used in the following publication:

Ferrari, R., Figueira, W.F., Pratchett, M.S. et al. 3D photogrammetry quantifies growth and external erosion of individual coral colonies and skeletons. Sci Rep 7, 16737 (2017). https://doi.org/10.1038/s41598-017-16408-z

3D model of a plate coral (2014)

3D model of a plate coral (2015)

Comparison heat map illustrating coral growth between 2014 and 2015

When conducting temporal analyses, one crucial consideration is the consistent placement of the scale bar within images. Shifting the scale bar between photos can compromise its reliability, requiring manual placement of markers for each image throughout analysis to ensure accurate scaling. In regard to the plate coral images from 2014 and 2015, the researchers positioned the scale bar directly on top of the plate coral in some images, while in others, it was moved between shots.

Why was model alignment necessary?

Aligning the models ensures that the measurements focus solely on the coral, excluding surrounding reef structures. This is critical for accurate volume and area calculations.

Why is cropping the models important?

Cropping the models prevents reef structures from being included in the comparison, ensuring that the analysis focuses exclusively on the coral.

Changes:

Change in Volume (∆V): +0.020092 m³
Change in Surface Area (∆SA): +0.693 m²

Ecological Implications:
The increase in both volume and surface area indicates coral growth.

Large Scale Reef Orthomosaic Photogrammetric Analyses

This dataset, collected in January 2016 at a reef slope near Heron Reef (23.26′S, 151.54′E), was gathered by Dr. Figueira’s team from the University of Sydney. Using a custom camera rig with three GoPro Hero4 cameras spaced 1 meter apart, divers conducted three overlapping passes, efficiently capturing a 300 m² reef area in just 7–10 minutes. Kindly note, this data was used in the following publication:

Lechene, M. A. A., Haberstroh, A. J., Byrne, M., Figueira, W., & Ferrari, R. (2019). Optimising sampling strategies in coral reefs using large-area mosaics. Remote Sensing, 11(24), 2907. https://www.mdpi.com/2072-4292/11/24/2907

Orthomosaic of Heron Reef Study Area

Photoquadrat illustrating 10 digitized table coral colonies, outlined with polygons (to enable spatial analysis and assessment of coral colony distribution and size)

Final Calculations

Area of live tissue in quadrat (m2): 1.412028

Area of mosaic (m2): 493.428

Area of quadrat (m2): 22.62

Percent cover of coral (%): 6.242387268

Coral density (per m2): 0.442086649

Avg distance (nearest neighbour) (m): 0.3503714286

Est. total area of live coral tissue for entire orthomosaic (m2): 30.801687

Page updated

Google Sites

Report abuse