Available for an ArcGIS organization with the ArcGIS Reality license.
The parameters used in computing the block adjustment are defined in the Adjust window. The available adjustment options depend on the type of workspace you defined when you set up the Reality mapping project. For example, frame triangulation is performed for aerial images.
Block adjustment
The block adjustment parameters for digital aerial imagery are described below. These parameters are used when computing tie points or ground control points (GCPs) and when computing a block adjustment.
For information about adjustment options for drone imagery, see Adjustment options for Reality mapping drone imagery.
Perform camera calibration
Automatic camera calibration computes and improves the camera’s geometric parameters, including interior orientation and lens distortion, while determining image orientation and image ground coordinates. If the camera has not been calibrated, select the options below to improve the overall quality and accuracy of bundle block adjustment. You can calibrate your camera during block adjustment to improve the camera's parameter accuracy. Most high-quality digital cameras have been calibrated, in which case these options should not be checked. This is the default setting.
- Focal Length—Refines the focal length of the camera lens
- Principal Point—Refines the principal point of the autocollimation
- K1,K2,K3—Refines the radial distortion coefficients
- P1,P2—Refines the tangential distortion coefficients
- Fix Image Location for High Accuracy GPS—This option is used only for imagery acquired with high-accuracy, differential GPS, such as real time kinematic (RTK) or post processing kinematic (PPK). If this option is checked, the process will only adjust the orientation parameters of the imagery and leave GPS measurements fixed. Ground control points (GCPs) are not required when this option is checked.
For more information about calibration options, see Camera table schema.
Note:
Higher in-strip and cross-strip aerial image overlap is recommended for better block adjustment and product generation results.Blunder point threshold (in pixels)
Tie points with a residual error greater than the Blunder Point Threshold value are not used in computing the adjustment. The measurement unit of the residual is pixels.
Prior accuracy setting
The prior accuracy setting allows you to specify the accuracy of orientation data from position and orientation measurement systems (POS) such as Applanix. Different POS may provide different accuracy information. For example, one POS may only output positional accuracy, while another may provide accuracy in x, y, and z. As a result, you only need to input the accuracy information that is available. The default setting for this category is null.
Digital airborne platforms measure exterior orientation using a POS. You can provide the measured accuracy of these parameters to improve the quality of the adjustment.
| Image Location (meter) | Description |
|---|---|
X | The accuracy of the x-coordinate provided by the airborne POS. The units must match PerspectiveX. |
Y | The accuracy of the y-coordinate provided by the airborne POS. The units must match PerspectiveY. |
Z | The accuracy of the z-coordinate provided by the airborne POS. The units must match PerspectiveZ. |
Planar | The accuracy of the x,y coordinate provided by the airborne POS. The units must match PerspectiveX or PerspectiveY. |
Position | The accuracy of the x,y,z coordinate provided by the airborne POS. The units must match PerspectiveX, PerspectiveY, or PerspectiveZ. |
| Attitude (degrees) | Description |
|---|---|
Omega | The accuracy of the omega angle provided by the airborne POS. The units are in decimal degrees. |
Phi | The accuracy of the phi angle provided by the airborne POS. The units are in decimal degrees. |
Kappa | The accuracy of the kappa angle provided by the airborne POS. The units are in decimal degrees. |
GNSS settings
GNSS settings provide options for calibrating the offset between the GPS antenna, camera, and GPS signal global shift in bundle adjustment. To use the GNSS settings, GCPs must be included in the adjustment.
For some airborne acquisitions, the GPS antenna is located separately from the camera system. The following options allow you to correct offsets in position measurements due to the physical offset between the camera and GPS antenna:
- Compute Antenna Offset—Corrects errors in sensor position by computing the physical offset between a camera and airborne GPS antenna
- Compute Shift—Corrects for instrumental drift in the GPS signal
Compute posterior standard deviation for images and solution points
The following options allow you to compute the standard deviation for each image exterior orientation parameters and solution point coordinates:
- Compute Posterior Standard Deviation for Images—The posterior standard deviation of solution points after adjustment is computed. The computed standard deviation values are stored in the Solution table.
- Compute Posterior Standard Deviation for Solution Points—The posterior standard deviation of each image location and orientation after adjustment are computed. The computed standard deviation values are stored in the Solution Points table.
Reproject tie points
A part of the adjustment process includes computing and displaying each tie point at its correct 2D map location. This is an optional step that only supports the visual analysis of tie points with the 2D map view. Following adjustment, the Reproject Tie Points option in the Manage Tie Points drop-down menu must be used.
Note:
When working with large projects with over a 1000 images, this step can be skipped to reduce adjustment processing duration, without any adverse impact to the adjustment quality.
Tie point matching
Tie points are points that represent common objects or locations within the overlap areas between adjacent images. These points are used to improve geometric accuracy in the block adjustment. The Tie Point Matching category in the Adjust tool includes options to support the automatic computation of tie points from overlapping images. Check the Full Frame Pairwise Matching check box to enable the automatic computation of tie points. The following conditions must be met for optimal results:
- Topography imaged is highly variable, for example, hilly terrain with large variations in height.
- Forward and lateral overlap percentages between images are lower than the recommended value.
- The accuracy of the initial imagery orientation parameters and projection center coordinates are low.
- Images have high oblique angles.
Image location accuracy
The inherent positional accuracy of the imagery depends on the sensor viewing geometry, type of sensor, and level of processing. Positional accuracy is typically described as part of the imagery deliverable. Choose the keyword that best describes the accuracy of the imagery.
The values consist of three settings that are used in the tie point calculation algorithm to determine the number of images in the neighborhood to use. For example, when the accuracy is set to High, the algorithm uses a smaller neighborhood to identify matching features in the overlapping images.
| Setting | Description |
|---|---|
Low | Images have poor location accuracy and large errors in sensor orientation (rotation of more than 5 degrees). The scale invariant feature transform (SIFT) algorithm is used, which has a large pixel search range to support point matching computation. |
Medium | Images have moderate location accuracy and small errors in sensor orientation (rotation of less than 5 degrees). The Harris algorithm is used with a search range of approximately 800 pixels to support the point matching computation. This is the default setting. |
High | Images have high location accuracy and small errors in sensor orientation. This setting is suitable for satellite imagery and aerial imagery that has been provided with exterior orientation data. The Harris algorithm is used with a small search range to support point matching computation. |
Tie point similarity
Choose the tolerance level for matching tie points between image pairs.
| Setting | Description |
|---|---|
Low | The similarity tolerance for the matching imagery pairs is low. This setting produces the most matching pairs, but some of the matches may have a higher level of error. |
Medium | The similarity tolerance for the matching pairs is medium. This is the default setting. |
High | The similarity tolerance for the matching pairs is high. This setting produces the least number of matching pairs, but each matching pair has a lower level of error. |
Tie point density
Choose the relative number of tie points to be computed between image pairs.
| Setting | Description |
|---|---|
Low | The fewest number of tie points is produced. |
Medium | An intermediate number of tie points is produced. This is the default setting. |
High | A high number of tie points is produced. |
Tie point distribution
Choose whether the output points have a regular or random distribution.
- Random—Points are generated randomly. Randomly generated points are better for overlapping areas with irregular shapes. This is the default setting.
- Regular—Points are generated based on a fixed pattern.
Mask polygon features
Use a polygon feature class to exclude areas you do not want used when computing tie points.
In the attribute table of the feature class, the mask field controls the inclusion or exclusion of areas for computing tie points. A value of 1 indicates that the areas defined by the polygons (inside) are excluded from the computation. A value of 2 indicates that the areas defined by the polygons (inside) are included in the computation and areas outside of the polygons are excluded.
References
Lowe, David G. (1999). "Object recognition from local scale-invariant features". Proceedings of the International Conference on Computer Vision. Vol. 2. pp. 1150–1157.
Chris Harris and Mike Stephens (1988). "A Combined Corner and Edge Detector". Alvey Vision Conference. Vol. 15.