Structure from motion photogrammetry in ecology: Does the choice of software matter?

Abstract

Image‐based modeling, and more precisely, Structure from Motion (SfM) and Multi‐View Stereo (MVS), is emerging as a flexible, self‐service, remote sensing tool for generating fine‐grained digital surface models (DSMs) in the Earth sciences and ecology. However, drone‐based SfM + MVS applications have developed at a rapid pace over the past decade and there are now many software options available for data processing. Consequently, understanding of reproducibility issues caused by variations in software choice and their influence on data quality is relatively poorly understood. This understanding is crucial for the development of SfM + MVS if it is to fulfill a role as a new quantitative remote sensing tool to inform management frameworks and species conservation schemes. To address this knowledge gap, a lightweight multirotor drone carrying a Ricoh GR II consumer‐grade camera was used to capture replicate, centimeter‐resolution image datasets of a temperate, intensively managed grassland ecosystem. These data allowed the exploration of method reproducibility and the impact of SfM + MVS software choice on derived vegetation canopy height measurement accuracy. The quality of DSM height measurements derived from four different, yet widely used SfM‐MVS software—Photoscan, Pix4D, 3DFlow Zephyr, and MICMAC, was compared with in situ data captured on the same day as image capture. We used both traditional agronomic techniques for measuring sward height, and a high accuracy and precision differential GPS survey to generate independent measurements of the underlying ground surface elevation. Using the same replicate image dataset (n = 3) as input, we demonstrate that there are 1.7, 2.0, and 2.5 cm differences in RMSE (excluding one outlier) between the outputs from different SfM + MVS software using High, Medium, and Low quality settings, respectively. Furthermore, we show that there can be a significant difference, although of small overall magnitude between replicate image datasets (n = 3) processed using the same SfM + MVS software, following the same workflow, with a variance in RMSE of up to 1.3, 1.5, and 2.7 cm (excluding one outlier) for “High,” “Medium,” and “Low” quality settings, respectively. We conclude that SfM + MVS software choice does matter, although the differences between products processed using “High” and “Medium” quality settings are of small overall magnitude.

Species and habitat mapping in two dimensions and beyond. Structure-from-Motion Multi-View Stereo photogrammetry for the Conservation Community

“…Both have drawn attention from the scientific community, but with ODM being the newer option, the focus has primarily been on its development and initial quality testing (3,4,53,54). MicMac’s maturity has allowed for a deeper scrutiny (55,56), has been utilised as the primary SfM-MVS option in geomorphological studies (57) and has been shown to be comparable to both Pix4D and Agisoft Photoscan (Metashape) within a low sward grassland study (58).…”

“…NodeMicMac is still in its infancy but now provides a user two different SfM-MVS processing options from within the one, WebODM interface. Also, as the original version of MicMac can have a steep learning curve (58), this provides an easier, more user-friendly option for deploying MicMac.…”

“…It is important to know what question needs answering beforehand and what scale of error is acceptable. For example, if wanting to produce a fine spatial resolution map, or look at animal species relationship to habitat structure, absolute precision and accuracy might not be as important as it would be with wanting to establish plant below ground biomass derived from above ground biomass (13) or other volume critical measurements (58). Alongside the aforementioned, there are some more useful general principles for data collection which can be applied to any and all SfM-MVS projects; Collect data with as high an overlap as is practicable – Nearly every SfM-MVS software will have its own guidelines for the amount of side and forward overlap between images, these range from 60%/60% (63) to 60%/80% (64), though recent studies have shown that higher overlap, upto 90% in both side and forward, is better for accuracy and precision within SfM-MVS derived data (65).…”

“…Some SfM-MVS software can automatically correct for this extra distortion, but where not possible FOSS software such as Rawtherapee can be used to correct images before processing (32). Collect an extra data set – Forsmoo et al (58) suggest that, as part of any robust SfM-MVS workflow for ecological purposes, obtaining a replicate data set for comparison can help aid in understanding any error that may arise from the processing through an SfM-MVS pipeline. This could be achieved by collecting twice as many images than is required within the same data capture task. Practice and get to understand the software – from our experience, SfM-MVS software will output better or worse data based on slight changes, either in how data is collected, or in how options within the software are utilised.…”

“…Collect an extra data set – Forsmoo et al (58) suggest that, as part of any robust SfM-MVS workflow for ecological purposes, obtaining a replicate data set for comparison can help aid in understanding any error that may arise from the processing through an SfM-MVS pipeline. This could be achieved by collecting twice as many images than is required within the same data capture task.…”