Error
When processing a project with accurate geotags in WGS 84 coordinate system without using GCPs or when the GCPs coordinate system does not match the output coordinate system, a horizontal shift may occur between the reconstructed scene and ground truth/verification data.
Description
If a horizontal coordinate system transformation is required to project the image or the GCPs coordinate system to the output coordinate system, an offset could be introduced between the model and check points or GCPs, when these are added.
Pix4D acknowledges the requirement of a more accurate than the current coordinate system transformation for projects referring to the Swiss coordinate system (CH1903/LV95 and CH1903+/LV95) or to the British coordinate system (OSGB36). However, it is possible that there are more coordinate systems requiring grid corrections.
Cause
Horizontal grid corrections are not currently supported. The coordinate system transformations in Pix4Dmapper are based on the GDAL library. In some cases, a constant offset is applied while in reality, a more accurate transformation needs to be applied for some pairs of coordinate systems.
Solution
To ensure accurate results there are two workarounds:
1. Transform the image geotags to the output coordinate system before importing them. There are several online transformation tools, such as the REFRAME to transform CH1903/LV03 coordinates to CH1903+/LV95 and Ordnance Survey's coordinate transformation tool to transform OSGB36 coordinates to WGS 84. Ensure that the correct coordinate system is selected for the image geolocation.
2. Introduce GCPs measured in the output coordinate system of the project: How to include GCPs in the project. Pix4Dmapper assigns a higher weight to the GCPs and they will correct for any errors introduced due to the incorrect transformation.
Hi,
I have a P4RTK that I am using to capture survey data and process in Pix4D. The base station is positioned using a "here" shot, meaning a horizontal and vertical translation is required to bring the work into terms of the local coordinate system. I know what the translation parameters will be, and they are in the order of 1.3m horizontally and around 30m vertically.
It is written above that horizontal grid corrections are not currently supported. What are the implications of using the RTK accurate EXIF position data from the photos to get the model to process, then use just one GCP to translate the data horizontally and vertically?
Is this the best way to handle the translation issue?
Thanks
Rhys
Hi Rhys,
Please create a Community topic here.
We are closing the articles' comments section as we want to do all the troubleshooting via our Community. It permits us to share information more easily.
Thank you for your understanding.