Tie points are pairs of pixels corresponding to the same locations where images overlap. They are an integral part of the bundle adjustment and contribute to most of the relative accuracy in the orthorectified product.

Note: Be sure that your tie points are of a quality that is consistently better than the initial accuracy of the sensor model; this varies with data type and vendor. For example, mid-latitude QuickBird scenes of a relatively flat area may provide an accurate orthorectification without the use of tie points.

When you click Next in Step 2, the ENVI Orthorectification Wizard proceeds to the optional Tie Point Selection panel (Step 3 of 5) if you selected more than one image for input. Two display groups also appear.

The controls in this step of the wizard allow you to optionally restore tie point files, to add new tie points, and to edit existing tie points. You may choose to bypass this step by clicking Next. The orthorectified solution will be computed without tie points in this case.

See the following sections:

Tie Point File Format


Rigorous Orthorectification supports tie point files in the following format:

; ENVI Rigorous Orthorectification TiePoint File
; FileName0=image1.TIF; FileName1=image2.TIF
; Image1File#, Image2File#, Image1(ImageX, ImageY), Image2(ImageX, ImageY)
;    0    1 195.250000 370.250000 6562.750000 398.250000

The first five lines in this example contain header information that lists the image filenames and the column labels.

The header rows that say "FileNamex" refer to the input image filenames, where x is an index number used to reference the filename throughout the rest of the tie points file. In the example above, “0” refers to image1.TIF and “1” refers to image2.TIF, as indicated in the header.

195.250000 and 370.250000 are the ImageX and ImageY values for image1.TIF.

6562.750000 and 398.250000 are the ImageX and ImageY values for image2.TIF.

If you are creating your own tie point file in a text editor or other application, you do not need to include a full path to the image filenames (in the lines that begin with "FileNamex") as long as your filenames are unique. In cases where the filenames may be the same for all images (for example, Cartosat-1 and SPOT data), you need to include the full path to each image file in the "FileNamex" lines.

When you restore a tie point file, ENVI appends five new columns to the table displayed in the Orthorectification Tie Points List: Correlation, RMSE, DX, DY, and Magnitude.

Correlation and RMSE are image-to-image error metrics that are used to estimate the quality of a tie point match. These metrics are calculated for image chips that are the same size as the moving window size and that are centered on the estimated tie point image coordinates. (See Step 4 in Automatically Generate Tie Points for a description of moving window size and image chips).

DX, DY, and Magnitude show the difference (in meters) in model-predicted map coordinates for the two tie-point image coordinates.

The image error metrics and image-to-ground difference metrics can be used to determine potentially poor quality tie points.

Restore Tie Point Files


Follow these steps to restore an existing tie point file:

  1. Click the Load Tiepoints button. The Select Tiepoints File dialog appears.
  2. Select a tie points file, and click Open. The tie points are listed in the Orthorectification Tie Points List (click Show List in the ENVI Orthorectification Wizard), and their locations are displayed in the Layout View Area of the Layout Manager (if you enabled the Tie Point check box).

Manually Add Tie Points Between Two Images


Follow these steps to add new tie points to an existing set, or to create a new set of tie points between two images.

  1. From the Image 1 drop-down list, select a filename for the first image from the list of open files. Or, Shift-click an image boundary in the Layout Manager to select it as Image 1. The first display group updates to show the image you selected. Image 1 is shown with a red border in the Layout Manager.
  2. From the Image 2 drop-down list, select a filename for the second image from the list of open files. Or, Ctrl-click an image boundary in the Layout Manager to select it as Image 2. The second display group updates to show the image you selected. Image 2 is shown with a yellow border in the Layout Manager.
  3. See Set the Image Order for more information about reordering images.

  4. In the Image window corresponding to the first image, go to where you want to add a new tie point.
  5. Left-click on a specific pixel to center over that pixel or portion of that pixel.
  6. The X/Y fields next to the Image 1 field in the table at the bottom of the ENVI Orthorectification Wizard show the image coordinates of the selected pixel underlying the crosshairs in the Zoom window. Subpixel coordinates are supplied to provide higher accuracy in selecting tie points. Whole-number pixel coordinates correspond to the upper-left corner of the pixel. The x,y values increase to the right and bottom of the pixel, respectively.

  7. Identify the same point in the second image, and repeat Steps 3-4 to define the pixel coordinates for that point in the display group corresponding to the second image. The X/Y fields next to the Image 2 field in the ENVI Orthorectification Wizard show the image coordinates of the selected pixel.
  8. Note: You can also enter values directly into the Image 1 and Image 2 X/Y fields if you have already determined their image coordinates.

  9. Click Add Point in the ENVI Orthorectification Wizard.
  10. Click Show List to display the Orthorectification Tie Points List. The new tie point is added to this table.
  11. When you add a tie point, a marker is placed in the Layout View Area of the Layout Manager, which consists of an ID next to an X symbol. The marker indicates the selected pixel (or subpixel location). The center of the marker (located under the X symbol) indicates the tie point location.

  12. Add more tie points as needed, using the same procedures.
  13. Note: The RMSE value will update when you add tie points.

Automatically Generate Tie Points


Images must have rational polynomial coefficients (RPCs) for automatic tie point generation to work. If your images do not have RPCs, you will need to manually add tie points.

Rigorous Orthorectification uses an area-based image matching method, which compares the gray scale values of patches of two or more images and tries to find conjugate image locations based on similarity in those gray scale value patterns. Tie points are generated on the single-band image that you set in Options > Band Selection.

Follow these steps to automatically generate tie points between overlapping images.

  1. To generate tie points for two images, select an image filename from the Image 1 and Image 2 lists. The two images must be different. Enable the All Images option, and click Auto Generate Points.
  2. Note: A warning message will display if you select the same image from the Image 1 and Image 2 drop-down lists.

  3. To generate tie points for all images where there is at least a 50% overlap in one dimension (for example, along an image edge), enable the All Images and Majority Overlap options, and click Auto Generate Points.
  4. To generate tie points for all images, enable the All Images option only and click Auto Generate Points.
  5. An Auto Tie Points dialog shows the progress of generating tie points for each pair of overlapping images, and the tie points are displayed in the Layout View Area of the Layout Manager and in the ENVI display groups that show the active images.
  6. Tie points are generated based on a default set of parameters. To change these parameter values, select Options > Auto Tie Point Parameters from the ENVI Orthorectification Wizard menu bar. The Generate Tie Points dialog appears. The following parameters are used:
    • Number of Tie Points: Specify the number of tie points to generate. This can be as few as 1, but the recommended value is 6 (default setting).
    • Search Window Size: Specify the search window size, in square pixels. The search window is a defined subset of the image, within which the smaller moving window scans to find a topographic feature match for a tie point placement. The search window size can be any integer greater than or equal to 21, but it must be larger than the Moving Window Size. The default is 101. This value depends upon the accuracy of RPC information of the images, and it also depends on the roughness of terrain.
    • Moving Window Size: Specify the moving window size, in pixels. The moving window scans methodically in the image subset area defined by the Search Window Size, looking for matches to a topographic feature. The moving window size must be an odd integer. The smallest allowable value is 5. The default is 31. Using a larger value results in a more reliable tie point placement, but takes longer processing time; conversely, a smaller value takes less processing time, but the tie points are less reliable. Determining a good moving window size largely depends upon the image resolution and terrain type. Some general guidelines follow:
      1. For a 10 meters or higher resolution image, use a range of 9–15.
      2. For a 5–10 meter resolution image, use a range of 11–21.
      3. For a 1–5 meter resolution image, use a range of 15–41.
      4. For a 1 meter or less resolution image, use a range of 21–81 or higher.
      5. Because most buildings in very high-resolution images in an urban area appear similar, specify a larger value to take the surrounding area into account for more robust matching results. To reduce the computational time for image matching, ENVI automatically uses a hierarchical matching method when the moving window size is greater than or equal to 19.
    • Minimum Correlation: Specify a floating-point value for the minimum correlation coefficient. Any matches with a correlation coefficient smaller than this value are discarded. The default is 0.55.
    • Chip Size: Specify an integer value for one side of a square image chip size for finding the interest points in the base image. Larger values require more computational time; however, the obtained tie points may be more useful. The default is 128, meaning a 128 x 128 image chip size.

Edit Tie Points


You can edit the location of a tie point by changing the image coordinates in the ENVI Orthorectification Wizard, or by interactively selecting new locations in the Zoom window.

To manually edit the location of a tie point, perform the following steps:

  1. Click Show List to display the Orthorectification Tie Points List.
  2. Click a row number to select a given tie point. Both display groups center over the tie point location.
  3. Enter updated Image 1 and Image 2 X/Y values in the ENVI Orthorectification Wizard.
  4. Click Update in the Orthorectification Tie Points List. The updated tie point location is reflected in this table, in the display groups, and in the Layout Manager.

To interactively move a tie point, perform the following steps:

  1. Select the row for that tie point in the Orthorectification Tie Points List, or click in a table cell within that row and click Goto. Both display groups center over the tie point, and the tie point is highlighted in the Layout Manager. If both images have consistent georeferencing (usually the case for adjacent along-track images in the same pass), tie points in the Layout Manager will be closely spaced and will appear as a single point. Otherwise, they will be shown as two distinct points.
  2. In each display group, center the crosshairs in the Zoom window over a new pixel location for the tie point.
  3. In the Orthorectification Tie Points List, click Update. The changes are reflected in the table, the display groups, and the Layout Manager.

To see which tie points have the highest error magnitudes, scroll to the right until you see the Magnitude (m) column. From the Orthorectification Tie Points List menu bar, select Options > Order Points by Error.

Remove Tie Points


To remove a tie point, select the row for that tie point in the Orthorectification Tie Points List, and click Delete. The RMSE value will update when you delete any tie points.

Turn Tie Points On and Off


Each time you turn a tie point on or off, the model is recomputed and the RMSE updates. If there are no GCPs, then turning on or off a tie point has no effect on the RMSE, but it does affect the model.

  1. In the Orthorectification Tie Points List, click on a row corresponding to a tie point you want to turn off.
  2. Click On/Off. The color of the tie point marker in the display group changes to green (by default) to indicate that it is off. You can change the default colors by clicking the Set TiePoint Color button. The Point Color dialog appears. Right-click on a color box and select the desired color.

When you click Next to proceed to the Image Order and Cutline panel, tie points that were previously off are turned back on and used in the model. Tie points that you deleted will remain deleted. The purpose of the On/Off functionality is so that you can experiment with tie points and decide whether or not to permanently delete any.

Note: Restoring a project during Step 3 (tie point selection) means that all GCPs will be turned on but some tie points may still be turned off, depending on how you set them before saving the project.

Save Tie Point Files


To save edits to an existing tie points file, or to save tie points to a new file:

  1. Click the Save Tiepoints button in the ENVI Orthorectification Wizard.
  2. The Select TiePoint File dialog appears.
  3. Specify an output location and filename, then click Save.

Next Steps


Proceed to Reorder Images and Define Cutlines.