This tutorial outlines the workflow for processing UAV imagery using Geomatica OrthoEngine. This tutorial shows how to import data, automatically collect tie points, extract a DEM and create a mosaic.
In Geomatica Banff SP1 new functionality was added for the UAV workflow including automated extraction of camera information, camera self-calibration, 2-pass tie-point collection, and improved model refinement. You can also now extract a DEM and generate the final mosaic in the same step, simplifying the workflow.
For this tutorial we will be using a UAV dataset from senseFly of a quarry in Switzerland. This data was collected by a Canon IXUS 220HS camera. You can download the data from the following link: https://pcigeomatics.sharefile.com/d-s9c0c1c5827c458ab
To begin, you will need to create a new OrthoEngine project.
- Open OrthoEngine from the Geomatica toolbar.
- On the OrthoEngine toolbar navigate to File > New. The Project Information window will open.
- Give your project a Filename, Name and Description
- Select Aerial Photography as the Math Modeling Method
- Select Digital/UAV under Options
- Select Compute from GCPs and tie points under Exterior orientation
- Click OK
- The Digital/UAV Camera-Calibration Information window will appear.
OrthoEngine supports externally calculated camera calibration, EXIF tag camera calibration extracted from an image, and manual entry. At minimum, focal length and chip size must be entered to continue.
- Perform one of the following steps:
- To use externally calculated camera calibration, click Text File… and choose the supplied camera calibration file xml
- To extract the focal length and chip size from embedded EXIF tags, click Image EXIF… and select a photo that will be used in the project. The following EXIF tags are required for automatic computation of focal length and chip size:
- EXIF.FocalPlaneResolutionUnit or EXIF.ResolutionUnit
- If no external calibration or EXIF tags exist, you must enter the focal length and chip size manually. Refer to your camera manufacturers website for this information.
- In this tutorial we will be extracting the information from the EXIF tags. Click Image EXIF and select one of the JPEG image files.
- Click Open
- Click OK
- The Set Projection window will appear.
This panel will be automatically populated if camera calibration information was extracted from EXIF tags.
Otherwise you can enter the information manually or click Cancel. If no information was entered and Cancel clicked, a popup will appear to tell you that the projection information will be read in from the first image.
- Click OK
Now that the camera calibration is set, we can add the JPEG files to the project.
- On the OrthoEngine toolbar switch the processing step to Data Input
- Click Open a new or existing image.
- You can choose to either:
- Add Folder - Import all jpeg or tiff image within a folder
- Add Image – Choose specific images to import
- Click Add Folder and navigate to the location of the .jpeg images
- Select the folder with the .jpeg images
- When prompted, choose YES to import the files to PIX format
- Click OK in the Multiple File selection window
- The new pix files will now appear in the Open Image window.
- Click Close
Import Exterior-Orientation (EO)
The JPEG images previously added to OrthoEngine contain EXIF tags and OrthoEngine imports these tags as metadata. Therefore, the exterior orientation (EO) is computed automatically. The following EXIF tags are required for automatic computation of the EO:
In this tutorial the EO information is imported from the EXIF tags. You can check the imported EO information by clicking the Manually Edit Exterior Orientation Data button.
When the input imagery does not contain the correct EXIF tags or you already have good EO data that you would like to use in your project, you would need to import the exterior orientation data from a text file:
- Click the Import exterior orientation data from text file button
- Continue through the steps outlined in the Geomatica Help - Importing GPS/INS or exterior orientation data from a text file.
Collect Tie Points
We will now need to collect tie points in order to improve the math model and ensure that the images align correctly.
- On the OrthoEngine toolbar switch the processing step to GCP/TP Collection
- From the OrthoEngine toolbar, click the Automatic Tie Point Collection icon
- In this window ensure the Matching method is set to Two Pass: FBM & FFTPB. This is a new tie point collection method that was added in Geomatica Banff SP1.
- In this tutorial we will keep the default parameters for each of the number boxes.
- Check Perform camera self-calibration
Note: If you did not use an externally calculated camera calibration, it is highly recommended to check Perform camera self-calibration.
- For this tutorial leave Retain tie points from first pass unchecked.
Note: This option may be required if few points are retained after the Two Pass: FBM and FFTPB method completes, but it is not recommended for general use as the first pass tie points are meant for rough orientation.
- Click Collect Tie Points.
Note that this process will take some time as there are many images.
- Once the collection process is complete a pop-up window will appear to indicate the tie point collection completion.
- Click OK.
- Click Close on the Automatic Tie Point Collection window.
GCP Collection (Optional)
In this tutorial ground control points (GCPs) are not collected. However, when running this same process with your own datasets you can collect GCPs to improve georeferencing of the dataset. You would collect GCPs on the images using an existing georeferenced image. The automatic GCP collection process is fully outlined in the Airphoto Orthorectification tutorial.
From the OrthoEngine toolbar open the Residuals report icon.
You can view the collected TPs in this window. Note the low X and Y RMS of 0.5. Additional information on the Residual Report is available from the Help.
Generating Epipolar Images
The next step in the process will be to extract a DEM. We will first create the epipolar images and then run the DEM Extraction process.
- Change the OrthoEngine processing step to DEM From Stereo.
- Click on the Create Epipolar Image
- For Epipolar selection, leave the default - Optimum pairs.
- Select All channels for the Left Image and Right Image.
- Click Add Epipolar Pairs To Table.
- Change the Downsample factor to 2 – This parameter controls the number of pixels that are used from the raw images to generate the epipolar image. Changing this to 2 will increase the processing speed.
- Leave the Edge clip (%) as 5
- Click Save Setup – the pairs will be generated when DEM Extraction is run
- You can alternatively choose Generate Pairs to generate the epipolar pairs before running DEM Extraction.
- Click Close
The next step in the process is to Extract the DEM from the epipolar pairs.
- On the OrthoEngine toolbar, select Extract DEM automatically.
- Click Select All under the Stereo pairs table.
- Change the Pixel sampling interval to 1. Since we created the epipolars at 2:1, this will produce a DEM at 2:1 GSD.
- Check the Epipolar Tracking
- Check the Create geocoded DEM
- Set an Output file name for the geocoded DEM.
- Set the X and Y resolution to 0.21.
- Check Create Image Mosaic.
- Set an Output file name for the image mosaic.
A color balanced mosaic of the images will be generated in addition to the DEM.
* Note – This is a new option available starting in Geomatica Banff SP1. The mosaicking method used at this stage varies slightly from the typical mosaicking workflow outlined below. This mosaicking method will reduce building lean and feature duplication in the output mosaic. This method also removes the requirement to run orthorectification as well as mosaicking as two additional steps.
- Click Extract DEM.
- Click OK
- Click Close
- You can load the geocoded merged DEM and the image mosaic in Focus for viewing.
Geocoded Merged DEM
All the required processing steps are now complete. However, there are additional optional steps that you can perform. When working on your own datasets if required you can edit the DEM. You can check the DEM Editing tutorial for detailed information on DEM editing.
You can also continue to clip out a specific area of interest as outlined in the next step. If required you can continue with the optional steps to generate the orthos and produce a mosaic in the Mosaic Tool from the orthos.
Clipping Final Mosaic (Optional)
In this tutorial we are specifically interested in the quarry so we will clip out the area of interest. You can follow these same steps clip out the same AOI from the DEM. Provided in the data download is an AOI vector file. This will be used as the clipping layer.
- In Focus open Tools > Clipping/Subsetting
- Under Input make sure that File is the UAV mosaic.
- Under the Available Layers section check off all three mosaic channels.
- Under Output set an output file location and name.
- For the Define clip region switch the Definition Method to Select a clip layer
- Browse to the location of pix.
- Change the Bounds selection to Shape(s) Boundary. The shape of the actual AOI polygon will then be shown.
- Click Clip
- You can now load the clipped mosaic file into Focus to view the final mosaic of the AOI.
Generate Ortho Images (Optional)
If you require orthorectified images, you can continue onto the Ortho Generation step in OrthoEngine.
- Change the OrthoEngine processing step to Ortho Generation
- Click the Schedule ortho generation icon
- In the Ortho Image Production window click >> to add all the Available images to the Images to process
- Select DEM File and then Browse to the geocoded DEM that you generated in the previous step.
- In the DEM File pop-up window select 1 [32R] Merged DEM and then click OK
- Click Generate Orthos
Mosaic Orthos (Optional)
A mosaic was generated when the DEM was extracted, however if required you can generate a mosaic from the orthorectified images. You can check the full automatic mosaicking tutorial for additional information on preparing, editing and generating mosaics in the Mosaic Tool.
- Change the OrthoEngine processing step to Mosaic
- Choose the Mosaicking The Mosaic Tool is then opened and the ortho images are automatically loaded in the New Project Wizard – Source Images window.
- Click Next
- On the Mosaic Definition window set the Output file. A single mosaic file will be generated.
- In the Mosaic Preparation window under Compute Cutlines change the Constrain parameter to Auto.
- Check off Sort Images
- Leave the rest of the parameters as default.
- Click Generate Mosaic - You can also choose Generate Preview and then further edit the mosaicking options and colour balancing before generating the mosaic as outlined in the full mosaicking tutorial.
- The mosaic is shown in the Mosaic Preparation window.
- Click Finish