![]() To perform the
registration
The Register
interface allows two basic procedures: the Image Registration and the Vector Registration of an Infolayer. The Image
Registration can be used
for georeferencing and later insertion in the GRIB format image database
(image file with the .grb extension). The Vector Registration
performs the georeferencing of Infolayers that contain vector data and
that belong to any Project linked to the current Database. The registration function
is available in SPRING's main module, after the activation of a Database.
In the case of Image Registration it is not necessary to load a Project
when the control points are acquired in the Table or Keyboard
modes. Load a Project only if the control points were obtained with
the Screen mode. In the case of the vector registration of an
infolayer one must work with a loaded Project, that contains the data
used as a reference for registration (Screen mode). However, it is
also possible to change the acquisition mode to Table or Keyboard. After recording in a remote sensing image, it is important to evaluate, or validate, the positional quality of the recorded image for different sensors from different satellite platforms.
With the objective of
allowing a better understanding the registration operation was divided in the
following items:
How to select an Image for RegistrationIn the beginning of registration one must select an image file (GRIB format only). The control points that will be defined later will be stored in the image file itself.
NOTE: If no Project is loaded, it will be necessary to inform in
the window
"Projection" the projection and Earth model (datum) that will
be used to register the image. How to select an infolayer for the vector registrationThe vector registration starts with the selection of an infolayer. The control points that will be used in the vector registration will be saved in a separate file, with the extension *.pro.
Acquisition of Control PointsWith the image to be registered drawn on Display 5, the user is ready to start the acquisition of control points in the window "Image Registration". Point AcquisitionAs was mentioned above, the user can acquire the points in three different ways: using a map in the digitizing table ( Digitizer mode), using an already georeferenced infolayer (Screen mode), or informing the coordinates on the keyboard ( Keyboard mode). See each of these modes below: Digitizer Mode
NOTE: If the digitizer is not calibrated the system will present the message "The Digitizer was not calibrated"; thus, place the reference map on the digitizer and start the calibration procedure.
NOTE: After dragging the point to its final position on the image, the
one deemed correct, it will change into a yellow cross. Note that from the
third point on the subsequent points are placed very near to their exact
position, for the system already uses the previous points to calculate the
correct position of the point being acquired. Screen ModeNOTE: The acquisition of the control points via Screen is only
possible with a loaded project. Thus before performing the registration in the Screen
mode make sure to load a project from your database (
OBS: Note that in the Reference Coordinates sector in the "Image
Registration" dialog, the exact coordinates of the point acquired from
the infolayer of the loaded project are presented. Keyboard Mode
NOTE: The coordinates of points, typed in on the keyboard, should usually be acquired by topographic or geodetic means (GPS, for example). This type of input is useful when there is no availability of a cartographic base to register an image or vector data. NOTE: In this type of control point input it is not necessary to have a
project loaded, since the coordinates of the points will be typed on the
keyboard. It is enough to inform what projection and Earth model (datum) will
be used. Other operations on control pointsSee following other operations for the manipulation of control points on the image or vector data to be registered. Delete Control Point
NOTE: The point deleted is still visible in Display 5 until you click on Execute
- Draw ( Move the Control Point
NOTE: The point moved keeps its original position until you click on Execute
- Draw ( Change the name of the Control PointTo change the name of the control point:
To acquire a featureNOTE: NOTE: Feature collation and editing can only occur in Screen.
OBS: Note that in the Reference Coordinates sector in the
"Image
Registration"
dialog, the exact coordinates of the point acquired from
the infolayer of the loaded project are presented. Refining the position of a control pointWhen we use another image as a reference for the acquisition of points in the Screen mode, we can use a feature to refine the position of a point that has already been inserted in the image to be adjusted (image in Display 5). To refine the position of a point the system uses an area of 17 x 17 pixels on the reference image and of 65 x 65 pixels in the image to be adjusted, in order to calculate the position through the analysis of an autocorrelation matrix. The maximum value in this matrix is 0.7; above that the system moves the point to a new position. The refinement of a point depends on some parameters:
The figures below show a
point on the reference image and the same point both before and after the
refinement in its position. REFERENCE IMAGE IMAGE FOR ADJUSTMENT IMAGE WITH REFINED POINT
NOTE: If the point in the image is off the limits defined by the area of
65 x 65 pixels, or below the maximum value for the autocorrelation, the Refine
button will not cause any effect. Exporting to ASCIIAfter the acquisition of the control points the user has the option of saving the control points into a file of SPRING's ASCII format.
Change the Image for AdjustmentIf the band or color composition used as image for adjustment is not sufficiently clear for the acquisition of the points, the user can change the bands since the points that have been acquired are the same for the image file (*.grb).
Selecting the Points for RegistrationAfter the acquisition of control points the user is ready to define which points will really be used for the definition of the mapping equation. The window "Image Registration" can be used to define the best points, which are those with the smaller errors, considering the choice of the degree of the polynomial that will be used to register an image. To assess the mapping equation we recommend that 6 control points be acquired for a first degree polynomial (minimum of 3 added of 3 more to calculate the errors) and 10 control points for 2nd degree polynomials (minimum of 6, plus 4 more to calculate the errors).
As the points are selected the errors are calculated (in units of resolution, pixel,
for the case of image registration, and meters in the case of vector
registration) and automatically presented to the side of the control
points. The calculations are also performed for the points not selected, which
serve as test points, that is, they allow a non-biased assessment of the
registration results. Note that underneath the Polynomial Degree radio buttons the total error
(in pixels, for images, or in meters for vector data) of the Test Points
(points that were not selected) and the Control Points (points that were
actually used for registration). The user should note the value presented in Error of the Control Points,
for it should be used to control the desired precision. In urban areas, for
example, one could consider a 0.5 pixel error for a
Another parameter used in the error analysis is a function of the scale that one
is working. For a mapping in the 1:50.000 scale, for example, the acceptable
registration error is about
Repeat, whenever needed, the control and test point selection operation, until a satisfactory result is achieved. You could test with other points, other polynomial, or even correct the position of some point. Saving the Parameters of a RegistrationAfter choosing the best points for a registration and defining the mapping equation, all of these parameters should be saved for further use. In the case of an image registration, the control points and the mapping equation are saved in the image file itself (*.grb). In the case of a vector registration, these parameters are saved in another file with the extension *.pro. SPRING uses the Import GRIB function to insert a registered image into a project and benefits from the Mosaic function to remap a registered vector data into a project.
NOTE: If the user tries to close the point acquisition window without saving them the system will present the message : "Control points nor saved. Do you wish to save them? Yes / No"
After this step the image can be imported into a SPRING project (see " Registering an image"), provided its bounding box contains the geographic area of the image. In the same way, a registered vector data can be mapped into a SPRING project (see Mosaic of Infolayers).
Validating control points or features.After choosing the best points and features that will be used in the registry and defining the mapping equation, the validation parameters can be chosen. Note that the field in the interface will be active if there are dots and / or features. You can choose one parameter at a time or both together. A report will be displayed showing the results of various calculations and can be saved in a file (.txt).
![]() Importing GRIB (Images) files into a Project. To Execute a Mosaic. The Logical Sequence to perform the registration. About Digital Images. ![]() |