Orthophotography
A general orthorectification workflow includes the following tasks:
Step 1: Orientating and rectifying the image
Interior Orientation – The process of establishing the internal relationship that exists between rows and columns on the digital image and the corresponding physical measurements on the image is called interior orientation.
Suppose we have an aerial photo with us, but it is a hard copy. To make a soft version of it, we are required to get it scanned to a reasonably high resolution. However, scanning would rasterise the photograph and the original camera internal co-ordinates would be distorted and the cell and row values would be assigned to each entity of the scanned image. How could we reset the relationship the between image and camera model?
The first step would be so-called interior orientation which means establishing the relation between (1) the camera-internal coordinate system and (2) the pixel coordinate system. The first relation is attributed by the so called fiducial marks which are superimposed in the image and their nominal coordinates, usually provided in mm in the camera calibration certificate. In this certificate you will also find the calibrated focal length in mm. The next step will be measuring the fiducial marks by manually digitising them to establish the transformation between camera and pixel co-ordinates. After digitising the marks, the transformation coefficients are calculated using an affine transformation. This calculation is generally done by the image processing software used for stereo model generation.
Exterior Orientation – This process is used for the determination of the camera location and orientation of the camera similar to that at the time of the image acquisition. Exterior orientation relies on the concept of the collinearity condition and is based on the six (out of nine unknowns) unknown parameters of the colinearity equation X0, Y0 and Z0.
Activity
Look again at the colinearity equation. Considering the parameters required for exterior orientation, can you explain what elements will be required to solve the equation? Where does the parameter- information come from?
Extraction of these parameters for automatic orthorectification is possible using GPS, IMU (Inertial Measurement Unit) and a DEM through retracing the image to ground.
If there is no information from GPS and IMU systems, then 6 exterior orientation parameters are calculated using space resection, as described above, for which photocoordinates of 3 GCP points are required (so 6 equations to solve 6 unknowns) and where there are more than 3 GCPS available, a least squares solution is found.
Step 2: Height information input and absolute orientation to produce orthophoto
In this process the objects in the photograph are rectified or registered with respect to the ground coordinates. For this, a DEM is used for the differential rectification. The aerial photograph is reprojected on the modelled ground positions (Horn, 1989). Fundamentally the process of reprojection can be done in two ways: Forward Projection (Direct Method) and Backward projection (Indirect Method).
In the case of forward projerction, the pixels from the original image are projected on top of the DEM of the 3D model and the pixels’ object space coordinates are estimated. Then, the object space points are reprojected into the orthoimage. In backward projection, the object space X, Y coordinates related to every pixel of the final orthoimage are determined first. The height Z at a specific X, Y point is calculated from the DEM or the 3D model and then the X, Y, Z object space coordinates are again projected over the original image in order to acquirie the values for the orthoimage pixel (Okeke, 2009).
The following animation presents a pictorial representation of the processes discussed above.
Source: Schowengerdt, 2007.
This animation shows the steps in the orthorectification process: