PCI Geomatics Help Center

How can we help you today?

Optical Satellite Orthorectification - Banff

PCI Geomatics -

This tutorial outlines the optical satellite orthorectification process using OrthoEngine. This workflow will show how to import data, collect GCPs and tie points and then orthorectify the images. Kompsat-3 data of Toronto, ON will be used for this tutorial, however these same steps can be followed with other types of satellite data. 

The following steps are completed in this tutorial: 

Project Setup and Data Ingest

Set GCP/TP Accuracy

Collect GCPs

Collect Tie Points

Point Thinning and Refinement

Point Thinning

Point Refinement

Residual Report

Generate Ortho Images

Project Setup and Data Ingest

To begin, you will need to create a new OrthoEngine project and ingest your imagery.

  1. Open OrthoEngine from the Geomatica toolbar.
  2. On the OrthoEngine toolbar navigate to File > New. The Project Information window will open.
  3. Give your project a Filename, Name and Description
    • Select Optical Satellite Modelling as the Math Modeling Method
    • Select the appropriate modelling method under Options:
      • Toutin's Model is a rigorous model that compensates for known distortions to calculate the position and orientation of the sensor at the time of image acquisition. It is suitable for use with any optical satellite data, regardless of resolution, such as CARTOSAT, LANDSAT, or SPOT.
      • Rational Function (Extract from image) can be applied to any image data set. If RPCs are unavailable, they can be generated from the points stored in the GCP segment (see RFMODEL). The advantage of using Rational Function (Extract from image) is that you can apply it to images that have been modified, do not contain an orbit segment, or both. This math model is recommended for use with data sets in map orientation, image subsets, or those with no orbital information.
      • Rational Function (Compute from GCPs) can be applied to any image data set. If RPCs are unavailable, they can be generated from the points stored in the GCP segment (see RFMODEL). The advantage of using Rational Function (Compute from GCPs) is that you can apply it to images that have been modified, do not contain an orbit segment, or both. This math model is recommended for use with data sets in map orientation, image subsets, or those with no orbital information.
    • Click OK
  4. The Set Projection window will appear. Click Cancel on this window. A popup will appear to tell you that the projection information will be read in from the first image.

  1. On the OrthoEngine toolbar switch the processing step to GCP/TP Collection.
  2. Click Open a new or existing image. Since Kompsat-3 data is directly supported we do not have to ingest it first. You can check the supported sensor list to see which data can be directly opened in Geomatica.

  1. Click Add Image… when the Open Image window opens.
  2. Navigate to and select the key file for your sensor. For Kompsat-3 data you need to select the Aux.xml image (K3_20130501182829_05101_19871327_L1R_Aux.xml) from the File Selector window. For other types of data, you can search the Geomatica Help for the sensor name to determine which file you need to use to open the image.

  1. When prompted, choose to add the MS bands and import the data file to PIX.
  2. Also choose yes when asked to create overviews.
  3. Your new pix file will now appear in the Open Image window.

  1. Repeat steps 8-11 for the other image (K3_20130501182829_05101_19871328_L1R_Aux.xml)
  2. If you wish to pansharpen your imagery you can do that before orthorectifying. You would need to also add the PAN band for each image through the Open Image window. You can then Pansharpen through OrthoEngine (Tools > Merge/Pansharped Multispectral Image).

Set GCP/TP Accuracy

Prior to collecting GCPs and TPs you’ll want to verify and adjust the GCP/TP Accuracy settings for your project. Setting the proper accuracy values is very important. If your project consists of surveyed GCPs, the accuracy entered should be sub- metre. More information on setting the accuracy of GCPs and TPs is found in the Geomatica Help.

  1. On the OrthoEngine toolbar choose the Project or GCP/TP Collection Processing step. You can then select the GCP/TP Accuracy button.

 mceclip0.png

  1. Choose the images that you would like to adjust the accuracy for from the Selection list. 
  2. You can then optionally choose a preset accuracy based on the options in the Presets list.
  3. Under GCPs, TPs, or both, enter the value by which you want to set the accuracy of the points in the Pixel, Line, Elevation (Z), Easting (X), and Northing (Y) boxes, as applicable.

 mceclip1.png

Click Apply. The image or images are updated with the new accuracy values based on the item you selected in the Selection list.

Collect GCPs

Now that the images are imported, you can collect GCPs and TPs. The following section of the Geomatica Help outlines the minimum number of GCPs to collect for each type of OrthoEngine math model: https://www.pcigeomatics.com/geomatica-help/concepts/orthoengine_c/Chapter_55.html

  1. On the OrthoEngine toolbar click the Collect GCPs Automatically button

 

  1. Change the Control Source to Georeferenced Image and browse to the reference Image.
  2. Browse to the DEM file.
  3. If required Adjust the Search Options:
    • Sample source method
      • Susan: sample points are generated automatically using the SUSAN corner-detection algorithm
      • Grid:sample points are generated automatically using sample points distributed evenly across the overlap region
    • Number of GCPs per image - Type the number of points that you want to collect over each image
    • Matching method
      • FFTP (Fast Fourier Transform Phase matching): when two images have a relative shift between them, the result is a phase difference in the Fourier domain. FFTP determines the shift between images using this phase difference.
      • NCC (Normalized Cross Correlation): this method finds the relative shift between two images by finding the shift that produces the maximum cross-correlation coefficient of the gray values in the images. This is the default method.
    • Search Radius - Type a number to define the search radius around the kernel
    • acceptance score- enter a value from 0-1 which defines the minimum correlation score that will be considered a successful match
  4. Click Match GCPs.

  1. Once the procedure is complete the collected GCPs will be listed in the Collected GCPs panel
  2. Click Compute Model to view the residuals for each of the GCPs

  1. If required you can change the Search Options and recollect GCPs. You can also select which GCPs you would like to use from the list of collected GCPs.
  2. To remove a GCP from the model calculations simply click on the checkmark in the Use as GCP

  1. Once you are satisfied with the GCPs in the list, click Add GCPs to Project. You can rerun the GCP collection after adding to the project to collect more GCPs.
  2. Close the GCP Collection window.

Collect Tie Points

Since we have two images and we want to ensure that the two images match each other, we will collect tie points in order to further improve the model. In the last year PCI has been working to improve the accuracy and speed of tie-point collection by up to 20%.

  1. From the OrthoEngine toolbar, open the Automatic Tie Point Collection window.

mceclip3.png

  1. Browse to the DEM file.
  2. Leave the other parameters as defaults. Similar to the GCP collection you will likely have to change the collection options based on your dataset. Open the Geomatica Help to read about each of the options
  3. Click Collect Tie Points and set the values for each of the parameters
    • Tie points per area - enter the number of TPs to use over the image or overlap area
    • Trials per point - enter a number representing how many attempts to make for each point before giving up on the match
    • Min. acceptance score - enter a value from 0-1 which defines the minimum correlation score that will be considered a successful match
    • Search radius - Enter a number representing the size of the area on the target image and select a unit of measure
    • Input elevation - Set either constant elevation or select a DEM file 
    • Distribution pattern 
      • Entire Imagedistribute the TPs over the entire image and match each TP in all of the overlapping images.
      • Overlap Area - evenly distribute the points only in the overlap area between any pair of overlapping images
    • Search method
      • Susan - find candidates by running a corner-detection algorithm on the image.
      • Grid  - find candidates in a grid-like pattern. In the Edge margin distance box, enter the minimum distance you want, in pixels, between the edge of the image and the placement of the candidates. 
    • Matching channels - Enter the channel or channels to use during collection

mceclip2.png

  1. Click Collect Tie Points. A pop up will appear to tell you how many tie points were collected. More or less tie points will be collected based on how much overlap is available in your dataset.

  1. You can change the Options and collect more tie points if required.
  2. Close the Tie Point Collection Window.

Point Thinning and Refinement

After you have collected GCPs and TPs you can further refine them if required. Make sure to calculate the model (Processing Step > Model Calculations > Compute Model) before this step.

In Geomatica Banff a new Point Thinning and Refinement window has been added. This window allows you to automatically run thinning to remove redundant points and refinement to remove points with high RMS. 

mceclip5.png

mceclip4.png

Point Thinning

To set up point thinning, you select the type of points you want to thin, how you want to dispose of the points designated for removal, and then define the grid by which to divide the image into cells. By defining the grid, you can specify the how many points you want to retain in each cell.

If you want to Thin Points make sure that the Thin Points box is checked off. 

  1. Points - Select the type of points you want to thin: GCPs, TPs, or Both.
  2. Number of cells on short side of each image - Enter the number of cells on the short side of each image.
  3. Maximum points per cell  - Enter the number of points to retain in each cell.
  4. Thinning action:
    • Deactivate  - Designate removed points as inactive but does not delete the point. You can switch deactivated points back to active in the future.
    • Delete -  Permanently remove the points from the project.
  5. You can either choose to also Refine Points or skip the point refinement and run only point thinning. 

Point Refinement

To set up point refinement, you select the type of points you want to refine, the mode to use, how you want to dispose of the points desinated for removal, and then enter the maximum x and y residual per point.

 If you want to Refine Points make sure that the Refine Points box is checked off. 

  1. Refinement Mode:
    • Progressive - Reject points iteratively. Points with a very high RMS threshold will be removed. The threshold is reduced iteratively based on the maximum x and y residual you specify.
    • Direct - Reject points at a specific threshold. Points are rejected at precisely the maximum x and y residual you specify.
  2. Points - Select the type of points you want to thin: GCPs, TPs, or Both.
  3. Maximum x, y residual per GCP/TP - Enter a value into either or both boxes depending on your points option. This value represents the threshold at which to reject the relevant points.
  4. Deactivate images with insufficient TPs - Deactivates images that do not have enough TPs. You would then specify the threshold in the Maximum number of TPs box. 
  5. Refinement action:
    • Deactivate  - Designate removed points as inactive but does not delete the point. You can switch deactivated points back to active in the future.
    • Delete -  Permanently remove the points from the project.
  6. You can either choose to also Thin Points or skip the point thinning and run only point refinement. 

Residual Report

From the OrthoEngine toolbar open the Residuals report icon

The collected GCPs and TPs are displayed in this window. There are quite a few options in the window. Explanations of each of the Options are available in the help.

You can switch to look at the GCPs, TPs or Both.

When you click on one of the points in the table you can see the residual summary for each image.

If you have any points with very high residual errors you can remove them from the model using the Delete Point button

Once you are satisfied click Compute Model and then Close the window.

Generate Ortho Images

You can now generate the orthorectified images.

  1. On the OrthoEngine toolbar switch to the Ortho Generation step and select the Schedule Ortho Generation button.

 

  1. In the Ortho Image Production window select the images in the left panel and add them to the right panel.
  2. Browse to the output location for your ortho images.
  3. Select the DEM.
  4. Click Generate Orthos.

Once the ortho step is complete you will have two orthorectified images. You can use these images to create a mosaic. Please refer to our in depth mosaicking tutorials for more information: Automatic Mosaicking and Manual Mosaicking

Have more questions? Submit a request

Comments

Powered by Zendesk