Page 106 - Digital Analysis of Remotely Sensed Imagery

Overview of Remotely Sensed Data 77

• Finally, the desired detail to be resolved should be several
times larger than the calculated spatial resolution of scanning,
depending on the distinctiveness of the features and their
contrast with the surroundings.
Compared with purchasing digital data, scanning analog print
materials to obtain digital data is rather inexpensive. A print costs a
small fraction of its digital counterpart. Besides, the obtained digital
photographs can be set at various spatial resolutions, though not finer
than the detail level of the original print. Data of a finer resolution are
obtainable with the use of a larger DPI, but raw digital data can be
degraded only from a fine resolution to a coarse resolution. In other
words, their spatial resolution cannot be made finer than the original,
no matter what kind of resampling scheme is used. Scanning aerial
photographs can also take advantage of the high geometric fidelity of
frame photographs because they are obtained at the instant of opening
the camera’s shuttle. By comparison, satellite images may suffer from
more geometric distortions during scanning that takes at least tens of
microseconds to complete. Furthermore, the obtained digital data
have a spatially uniform resolution. Pixel size hardly varies across the
photograph, in sharp contrast to satellite scanning in which pixel size
could be severely compromised at a large off-nadir viewing perspective.
In spite of the above advantages, there are three disadvantages
associated with scanning aerial photographs:
• First, photographs have a limited spectral range. Since films
can capture visible light and NIR radiation over the wavelength
of 0.4 to 0.9 μm, it is impossible to obtain data over the mid-
infrared or TIR portion of the spectrum.
• Second, scanned photographs can be separated only into three
layers: blue, green, and red. It is impossible to obtain more
spectral bands than this number. Besides, the exact spectral
range of each separated layer is not precisely known.
• Finally, artificial radiometric variations are inherent in one
photograph and across multiple photographs. It may not
be possible to eliminate the radiometric variation of the same
ground object across multiple scanned photographs.
The issue of artificial variation in radiometry over an aerial
photograph is usually dealt with by tinting the camera’s lens. The
nonuniformity in illumination caused by the absorption of a concave
lens is reduced to such a level that it is not a primary concern any
more. By comparison, the issue of varied radiometry across multiple
photographs is much more severe and difficult to tackle (Fig. 2.14).
Pseudoradiometric properties of the same ground feature across
photographs are produced out of two processes: