Download PDF
Download page Photogrammetry.
Photogrammetry
Introduction
Qinertia includes a full featured photogrammetry module that is designed to ease post processed direct georeferencing photogrammetry tasks. With this module, you can create, edit and review the match between pictures and events to update pictures EXIF/XMP metadata with precise position and orientation data. You can also export directly text based files that are supported directly in major photogrammetry software packages such as Pix4D, UAS Master, Agisoft Metashape, ...
The module has been designed to support both GNSS only and full INS solution and also includes a native support for DJI RTK capable products such as the DJI Phantom 4 RTK UAV.
Module Presentation
The photogrammetry module configuration is opened with the main tool bar icon and from the Project→Photogrammetry menu. From this dialog box, you can apply a shutter speed delay, configure the mechanical installation parameters such as the lever arm and alignment between the IMU and the camera CMOS sensor and review/edit the match between pictures and time events.
The screenshot below shows the photogrammetry main configuration dialog box for an INS project.
Camera settings
You can apply a constant time offset on all time tagged pictures to compensate for mechanical or electrical delays. The delay is added to the event time so a negative delay means the photo has been taken before the reported/measured time.
Qinertia can also apply an automatic delay based on the picture shutter speed. For some cameras it could be more accurate to always use a delay that is equal to half the shutter speed. You should of course refer to the camera specification and how the shutter events are timestamped.
Camera mechanical installation
If you are using an INS to update pictures with both an accurate position and orientation, you have to enter accurate lever arm and misalignment information between the IMU and the camera. This is mandatory to enable precise direct georeferencing and any inaccuracies in these settings will directly decrease the overall point cloud accuracy.
In Qinertia the misalignment between the IMU and the camera has been split into two alignments:
- The camera rotation that is mainly used to compensate for large misalignment between the IMU and the camera such as a 10° rotation. This is typically measured during payload design by a CAD software or directly during manufacturing and integration.
- The boresight angles that is an additional small roll, pitch and yaw misalignment to apply on top of the previous camera rotation. This small misalignment is generally computed during a dedicated calibration run and using a photogrammetry software.
The typical workflow, for a new payload, would be to enter the camera rotation and export the updated JPEG with precise orientation and position data. Then use a photogrammetry software to compute residual boresight angles during a dedicated calibration run with a lot of overlap and ground control points. Finally, you just have to enter the boresight angles estimated by the photogrammetry software in Qinertia. You can then re-export the JPEG to validate in the photogrammetry software there is no more boresight angle offsets.
GNSS only
For GNSS only and DJI projects, these settings are not available. Indeed, an orientation is needed to apply a lever arm expressed in body frame and convert it to navigation frame (North East Down). If you have a GNSS only project, the pictures will be updated with the position at the GNSS antenna unless you provide for each picture a North East Down offset to apply from the antenna to the camera as it is done by DJI in MRK files.
Pictures & event match
An important part is to correctly match each picture with a time to update it with the correct position and orientation. This task is done using a dedicated match editor to ease this job as much as possible. Simply click on the Open Editor button to show it and refer to its documentation below.
Pictures Match Editor
The match editor lets you create, review and edit the match between an event (time) and a picture. This is important so Qinertia can correctly update each picture with exact position and orientation. There are mainly two ways to create a picture / time match:
- Load a list of time markers events and a list of pictures to match each event to each picture
- Load directly a match file that associate a picture with an accurate time
Load a match
If you are using an integrated camera payloads such as DJI Phantom 4 RTK UAV, a specific match file might be available. For DJI this file has a MRK extension and stores for each picture the exact time when the picture has been taken. You can directly load this file using the Import Match and Qinertia will display the time with associated pictures. Qinertia can also load and export generic match files as describe in the section Match File.
The picture below shows a Phantom 4 RTK DJI acquisition that has been loaded. Qinertia has successfully read from the MRK file, the list of events, pictures and APC to CMOS lever arm offsets. To work, the pictures have to be located in the same directory as the MRK file.
APC to CMOS offset
Qinertia can load from a match file a North, East, Down offset to apply to relocate PPK solution computed at the antenna phase center (APC) to the camera CMOS sensor. This is offset is only applied for GNSS only computations and it is used for example with DJI UAV.
If Qinertia has loaded an APC to CMOS offset, it is indicated by the icon and you can view the offset value in the picture details table.
Create your own match
If you are using an INS, you will generally connect the camera shutter trigger to an input event marker on the INS. For example, SBG Systems INS feature several event inputs that are loaded by Qinertia to precisely timestamp each camera shutter trigger. You then have to load this list of events as well as the list of pictures to create a match between an event and a picture. Simply use the Import Events button to either import events present in the project or load an external events file.
It is quite common to have some discrepancies between the events and the list of pictures and Qinertia offer a very intuitive tool to review and correct events / pictures match. From this panel, you can exclude any event or picture from the match to fix errors. You can also use the auto match feature by clicking the button. The auto match try to find the best solution based on delta time pattern analysis between events time and pictures times.
Using the editor
Once you have created a match or imported one, you should see the following dialog box. On the left you have the list of events / time and on the right the list of pictures. In the following example, we have a perfect match with one event for one picture.
In the example below, we have unselected both the event and the associated picture to exclude this picture form the project. This picture will not be exported and it can be useful if you don't want to include it in the photogrammetry processing. The rest of events/pictures match is left unchanged.
In the following example, a picture is missing and we see that an event has been skipped to fix this and get a 100% match. In this case, we have just used to auto match feature and Qinertia has detected the missing picture.
Pictures Exporter
This dedicated exporter is used to update pictures with precise orientation and position. This is the last step before you can import the pictures or the CSV compatible geolocation file into your photogrammetry software. Please refer to Export Module to get a full documentation on the exporter module it self.