ire
PHOTODIGITIZING
A designated area of a continuous-tone photograph is scanned in a recti-
linear fashion so that the area is completely covered by suitably sized
pixels (Fig 3). «The light intensity is integrated over each. pixel area
and a single numerical value is assigned to each pixel proportional to the
integrated photon flux. A "numerical image" is thereby created, with
spatially-distributed numbers representing the original scene in the input
photo. This numerical image is then encoded for storage on a suitable
medium, usually magnetic tape.
A corollary function of most photodigitizers is that of photographic play-
back (Fig. 4). Here, the sequence proceeds from the stored data, through
the numerical image, to the modulation of a light source which exposes
corresponding pixels on a photographic emulsion. No sub-pixel modulation
exists, and the resulting playback should not be confused with a continuous-
tone photograph; rather, it is a visual representation of numerical data.
Each pixel has a unique address and value: the x and y coordinates, and
opacity. Each of these numbers could be changed, individually or collect-
ively, while in the digital domain. Thus, an address change would represent
ageometrical transformation, while the pixel value could be modified as part
of a photometric enhancement routine (Fig. 5).
DIFFERENTIAL RECTIFICATION
The feature displacements mentioned earlier can each be compensated for
mathematically. Distortions caused by the camera lens and by film shrinkage
and lack of flatness can be modeled independently of the orientation of the
aerial photograph relative to the ground coordinates of the scene depicted.
The relief displacements, however, require an absolute orientation of the
photograph before corrections can be computed. The elements of this orienta-
tion include Tilt, Swing, Azimuth, Focal Length, Flying Height, Photo Nadir
Coordinates, and Ground Nadir Coordinates. If these elements are unknown,
they can be calculated by a process of analytic spatial resection involving
the measured photo coordinates of known Ground Control Points. The vertical
line intersecting the photograph at the nadir point and the terrain at the
Ground Nadir Point is used as an element which is common to both the photo
and the terrain.
In order to compute terrain relief displacement, a Digital Terrain Model is
required. This is a set of terrain elevations which is usually configured
in a regular grid pattern. The ground coordinates of the elevations must be
known so that the spatial relationship between these elevations and the
vertical nadir line can be determined. If the spatial density of elevation
information is insufficient to provide an elevation for each orthophoto
pixel, a scheme of interpolation for such data must be implemented.
In order to accomplish a digital differential rectification of an aerial
photograph, the pixel size and the area and location of the final orthophoto
product must be established. A known relationship must exist between the
orthophoto coordinates and the nadir line. This vertical nadir line ties
together the aerial photograph, the Digital Terrain Model, and the ortho-
photo. In addition, it ties each of these systems to a Ground Coordinate
System.
61