Tie points (GCPs, MTPs, and CPs) - PIX4Dmatic

GCPs and MTPs are used to improve the absolute and relative accuracy of photogrammetric projects while checkpoints are used for quality assessment

Description

• Ground Control Points (GCPs) are points with known coordinates. Typically, GCP coordinates are very precise and are measured using RTK/PPK GNSS receivers or total stations. GCPs are used to accurately georeference a project to assess the accuracy of the reconstruction.
• Manual Tie Points (MTPs) are 3D points corresponding to features that are marked (clicked) by the user in the images. They are used for assessing and improving reconstruction accuracy.
• Checkpoints are used to assess the absolute accuracy of the model and are not used for georeferencing the project.

In this article, GCPs, MTPs, and Checkpoints will be collectively discussed as Control and tie points.

Tie points can be imported and marked at different stages:

• Before the Calibration step: This requires more manual work as the position of tie Points on images is not precise as it is determined only based on the initial image geolocation and orientation. However, marking the tie points before processing can eliminate the need for reprocessing or running the Reoptimization step.
• After the Calibration step: The position of tie points on images is more precise as external and internal camera parameters are calculated. However, reprocessing or running Reoptimization is required.

Import tie points

It is possible to import GCPs and checkpoints in the .txt or .csv file format. For more information see Tie points (GCPs, CPs) import format - PIX4Dmatic.

To import tie points geolocation information:

• Click Tie Points to open the Tie points table.
• Drag and drop, or use the Select from disk option to specify the file with GCPs.
• Define the Horizontal and vertical coordinate reference system.
• (Optional) Define the Geoid or Geoid height.
• Click Apply.

After the tie points are imported, they are displayed in the tie points table, the 2D and 3D view.

Import and export marks (Optional)

Import marks

It is possible to import marks from PIX4Dmatic and PIX4Dmapper projects. This way it is not necessary to mark them in the images again. The .txt/.csv file should follow this format:

image_name,GCP1,3830.31,123.113,1.22678

Export marks

Add Manual tie points

There are three different ways to add MTPs to the project:

• [1] In the Image viewer, click Mark new MTP, or
• [2] In the Image viewer, right-click the image and click Add & and mark new MTP, or
• [3] In the Tie points panel, click and Add tie point.

Mark tie points

Once tie points are imported or added, they can be marked either before running the processing or after.

To do so:

1. In the Tie Points panel, click the tie point, or select the point in the 3D view.
2. On images, use left-click to mark the tie point. A yellow cross - appears each time the tie point is marked.
   
   Important: To change the display of an individual image:

   • Hold Ctrl and left-click to pan on the image.
   • Hold Ctrl and rotate the scroll wheel to zoom in and out of the image.

   For more information: Navigation and panel.
3. (Optional) Repeat steps 1 and 2 for the rest of the tie points.

Recommended workflow to mark tie points in PIX4Dmatic

The recommended workflow for efficient marking and use of the tie points in PIX4Dmatic is: 

• Import images
• Import GCPs or tie points
• Mark each tie point in 2-3 images
• Run Calibration step
• When successful, proceed with marking in more images, or using the Automark feature.
• After the new marks are completed, run Reoptimization

Automatic target detection - AutoGCP algorithm

AutoGCPs algorithm automatically locates targets in images and detects their centers with pixel-level accuracy. It reduces the time needed for tie points marking.

The recommended workflow to use the AutoGCP feature is:

• Import Ground Control Points (GCPs) and set the correct Coordinate Reference System (CRS).
• Calibration: Run the Calibration step with the preferred settings.
• AutoGCP marking: Run this function manually to optimize GCP placement.
• Check Marks: Review the check marks for GCP alignment accuracy. Manually adjust, add, or remove GCPs if needed.
• Reoptimize and Processing steps: If Calibration is successful and the alignment is correct, run the Repoptimization first. If successful, proceed with the next processing steps (Dense point cloud, Mesh, DSM, Orthomosaic).
• Quality Check: Conduct a quality check of your model. Export the Quality report to verify the accuracy, check for any misalignments or errors, and ensure that the results meet project specifications. 

Automatic marking, Auto-mark algorithm

The Automatic marking (Auto-mark) algorithm is used to automatically propagate marks on images. It is available when:

• The Calibration step is completed.
• At least two marks are added manually on images.

The algorithm searches for automatic color correlation of a feature on the rest of the images. Thereby, the position of the marked feature will be optimized in additional images if the color correlation is good. The images that have a yellow cross are taken into account for processing during the Calibration step.

Vertex tie points

The vertex tie points feature provides a method to improve project calibration incrementally by converting geometry vertices into tie points. The recommended workflow for vertex tie points is:

1. Initial Calibration: Run the Calibration step with the desired settings
2. Marking geometry vertices: Create geometries and mark vertices in the image viewer.
3. Converting to Vertex tie points: Marked vertices can be transformed into vertex tie points, which are then used to refine the Calibration. The conversion methods are:
   1. Automatic Conversion: Run the Reoptimization or Calibration processing steps with the Vertex conversion toggle enabled to convert multiple vertices at once.
   2. Manual Selection: Right-click on selected geometry vertices and select Convert vertices to tie points from the menu.
      
      
      

When using the vertex conversion, the newly created vertex tie points are available in the tie points table in PIX4Dmatic (incl. the tie points section in the Quality Report). 

Use tie points for processing

It is recommended to add the tie points before processing the project. If GCPs and checkpoints with a different coordinate system are imported after processing, the images will need to be reprocessed and previous calculations will be lost.

To process the project:

• Click Process or,
• On the Menu bar, click Process > Calibrate.

Checking tie points errors

Introducing tie points in the project gives insights into the accuracy of the reconstruction:

• GCPs and checkpoints are used to assess absolute accuracy.
• MTPs are used to assess the relative accuracy.

For more information: What is the relative and absolute accuracy of drone mapping.

Tie points panel

In the Tie points panel, the following can be used to assess the quality of the reconstruction:

• Reprojection error [px], the reprojection error of the tie point on images.
• Position error X [units], The difference between the computed and the initial position of the ground control point in the X-axis, i.e., initial position - computed position.
• Position error Y [units], The difference between the computed and the initial position of the ground control point in the Y-axis, i.e., initial position - computed position.
• Position error Z [units], The difference between the computed and the initial position of the ground control point in the Z-axis, i.e., initial position - computed position.

Outliers

When a tie point does not fit within the calculated position it is then considered an outlier. In PIX4Dmatic, this is indicated by a red circle in the tie points table found in the Status center.

Quality report panel

The following is available after running the Calibration or Reoptimization step:

• Detailed quality report (.pdf):
  • In the Reports panel, check the Quality Report box, then click on Export
  • Navigate to > File, click Export > Quality report...
• General quality report :
  • On the right bar, click Report