relief,leads to an improved capability of image enhancement by presenting
object characteristics without the obscuring effect of shadows caused by
relief (Horn et al., 1977). Kahle et al. (1977) had to eliminate the slope
, angle effects from thermal inertia images (Fig. 5).
(c)
Figure 5: Thermal inertia images and correction of slope effects (Kahle et al.,
1977): (a) raw image; (b) slope corrected; (c) synthetic slope image.
Aninteresting demonstration of the capabilities of image registration
as & pre-processing step was reported by Anuta et al., (1976). A total of
20 images (Landsat, radar, photography, non-natural (synthetic) images
such as terrain relief and geophysical parameters) were registered into
a data set with 20 channels. Further processing aims then at the compres-
sion and transformation for subsequent interpretation. This matching of
non-correlated imagery in a wide range of the electro-magnetic spectrum
with data bands employing geographically denoted cells, may become an im-
portant field of development in the near future.
4.2 Data Compression and Enhancement
Photo-interpretation is of records that are spatially two- or - three-
dimensional (stereo). For reasons of redundancy and cost, interpretation
of records of many spectral dimensions is not done, but for limited test
areas. Therefore, there is a need to compress multi-dimensional images
spectrally to a format of least dimensions.
tect However, some of the techniques applied to compression can also serve
to emphasize specific information (enhancement) in images that are al-
ready of limited dimensionality. The most important techniques used are:
principal component transformation(pcT),r&tioing, differencing, direction
consine transformation.
4.2.1 Principal Component and Canoncial Transformation
This is the most commonly used technique of compressing and enhancing
multiple, geometrically registered images. Given are n values x,, X55 X4»
Xo per pixel, where x. may be a spectral density (n perfectly redistired
imagés). PCT results in & new ordered set of n images called first, second