Page 196 - Digital Analysis of Remotely Sensed Imagery

Image Geometric Rectification 161

the use of GPS technology, the choice in selecting quality GCP candi-
dates is broadened.
Nevertheless, this superior performance of GPS technology is not
without limitations. Since coordinates have to be logged in the field, it
can take a lengthy time to travel there. Extended field trips are essential
in reaching distant points. Besides, the technology is highly restricted
by site accessibility. GCPs located in remote, isolated, or mountainous
areas are not so easily accessible if there is a lack of vehicle navigable
roads, such as in rural environments in some developing countries.
Accessibility is also an issue if GCPs are on private land. Prior access
authorization must be gained from respective land owners. This expen-
sive method is the only choice for areas where no up-to-date topo-
graphic map is available or where the ground has changed consider-
ably since the topographic map was compiled.

5.4 Rectification Models
The transformation of a local image coordinate system to a global one
requires a rigorous geometric model. Many geometric models have
been devised for f and f . Ranging from affine to projective transfor-
1 2
mation, these models are designed for processing images acquired
from a variety of sensors and platforms, and for areas of varying top-
ographic relief. Some of them are applicable to images obtained from
a particular sensor, whereas other generic ones can be used for all
types of imagery.

5.4.1 Affine Model
The affine model is a custom geometric correction model that allows
three modifications to be made to the input image: scaling (change in
pixel size), offset (lateral shift in image origin), and rotation (change
in image orientation). Through offset an image is moved laterally by
a user-specified number of pixels in both the easting and northing
directions. It does not involve any change in image geometry (e.g.,
shape and dimension).

5.4.2 Sensor-Specific Models
Dissimilar to physical models that are based on the imaging process
(Okeke, 2006), sensor-specific models depict the mathematic relation-
ship between the image space and the object space. As a function of
time over the imaging period, these models depict the position of the
satellite at the time of imaging They incorporate exterior orientation
parameters of the sensor supplied with the data in the transforma-
tion, such as the parameters accompanying the Landsat series of sat-
ellite data. Landsat-specific models can be used to rectify Landsat
Thematic Mapper (TM) and Multispectral Scanner (MSS) images