In this case study, we incorporated six GCPs into the ShapeMetriX georeferencing workflow, obtaining a positional accuracy of 0.016 m – taking performance from survey‑grade requirement to engineering‑grade certainty.
Step-by-Step Georeferencing Using GCPs
This case study demonstrates the process of achieving high-accuracy georeferencing in a 3D model generated from aerial imagery, using a single-bench open pit mine slope as an example. The goal: improving positional accuracy of the model by incorporating GCPs into the ShapeMetriX workflow.
Check out Essential Tips for Using Ground Control Points in ShapeMetriX article for critical insights about GCPs.
Step 1: Aerial Images
The example image dataset includes 33 overlapping aerial images captured in 4 rows, following the minimum 80% overlap and image quality recommendations outlined in the ShapeMetriX Field Procedures user guide. It also contains 6 GCPs for georeferencing our model to a higher-precision positional control.
As illustrated below, begin the 3D model generation process by loading the images into MultiPhoto for the Coarse Reconstruction step.
Step 2: Coarse Reconstruction
This step determines the relative positions and orientations of the cameras, generating a basic 3D point cloud outlining the scene’s structure.
The images are processed and aligned in coarse reconstruction, resulting in the placement of the green pyramids shown in Figure 2. These green pyramids represent the reconstructed camera positions, derived from the image data.
Next, enable the GPS positions stored in the image data by selecting Show Referencing Positions. The green spheres (GPS positions) should align with the reconstructed green pyramids (camera positions), as shown in Figure 3.
Note the discrepancies between the camera positions and the GPS positions. These offsets, called GPS residuals, highlight the differences between measured and calculated positions. Larger residuals indicate a greater mismatch between the measured and calculated positions usually indicating inconsistencies in the GPS data.
In this dataset, the misalignment (known as GPS residuals) averaged 1.09 m, highlighting the limitations of standard drone GPS for geotechnical integrity. Reconstruction statistics (mean error, maximum error, standard deviation) are automatically summarized in the ShapeMetriX Reconstruction Report after dense reconstruction.
Review the coarse 3D model and proceed to dense reconstruction.
Step 3: Dense Reconstruction
Proceed with the default settings to begin the dense reconstruction. To learn more about how dense reconstruction settings influence the resolution and level of detail in the resulting 3D model, refer to our article: Dense Reconstruction Presets and their Influence on 3D Model Quality.
Step 4: Referencing
ShapeMetriX automatically generates the dense reconstructed 3D model. In the final step, we will use externally surveyed GCPs to correct standard drone GPS residual errors to achieve centimeter-level accuracy in our georeferenced 3D model.
4.1 Load GCPs
Use Load Ground Control Points from Text File in MultiPhoto Referencing mode to import GCPs from a csv file.
4.2. Locate GCPs
Activate one GCP from the list and localize its position in the 3D viewer. This will update the list of reference images including specified position.
Locate the target discs (GCPs) in the images. Each GCP should be marked in at least two images, though three or more are recommended for greater stability.
When the Baseline and Inliers status signals turn green for the selected GCP, activate the next GCP and repeat the same steps until all GCPs are localized.
4.3. Complete Referencing
Referencing statistics are displayed interactively under the Status tab. After all GCPs are localized, the estimated mean positioning error (residual) is now 0.016 m.
By utilizing GCPs in the georeferencing workflow, standard drone GPS residuals are corrected to achieve centimeter-level accuracy in our 3D model. Save and exit the model.
The Verdict
Using six GCP to tightening positional accuracy just over a centimeter (0.016 m), ShapeMetriX transforms drone-derived imagery into survey-grade 3D models. This level of fidelity not only meets but exceeds the accuracy requirements for structural geological mapping, surveying, and geotechnical analysis.
The implications are direct: safer slope designs, more reliable rock mass characterizations, and greater confidence in every mapped discontinuity. Put simply: when errors are reduced from meters to centimeters, every decision is grounded in reality.
Key Takeaways
- Standard drone GPS typically drifts from real-world coordinates by meter-level: this may only be applicable for geotechnical tasks with limited demands on accuracy and interoperability.
- Incorporating GCPs into ShapeMetriX workflows corrects residual errors, achieving centimeter-level accuracy.
- Centimeter-scale precision supports safer engineering design, more dependable analyses, and geospatial models that decision-makers can trust.
