Research Methodology - Historical Land Use/Land Cover Classification Using Remote Sensing - page 79

A radiometric correction process was fulfilled on the mosaic scenes which
carefully covered the study area. This was achieved by accrediting one of the
scenes as a radiometric-reference (master-scene). Then, the other image/s were
matched with it radio-metrically, i.e., a transformation process of the radiometric
characteristics of the source-scene was conducted on the other images (targets).
This resulted in obtaining close and similar radiometric characteristics for all
scenes that covered the study area, because all had the same reference/source (i.e.,
the master-scene). Consequently, the Earth features (e.g., wheat fields) that existed
in an individual scene, appeared spectrally (reflectance values/gray values) and
radio-metrically, similar to those wheat fields located in each of the other scenes.
This degree of similarity was based on the applied radiometric correction method/s
and on the nature of the ground surface features that existed in the satellite image.
After finishing the atmospheric correction using ATCOR-2, a radiometric
correction process was conducted of the scenes covering the study area (MSS-
June-1975 and TM-August-2007) using iMAD. A radiometric correction was
applied upon the two mosaic-scenes, since the TM-data was too basic for use with
iMAD. As to the scenes that could not pass the radiometric correction process (for
instance, TM-May-2007-data), it was enough to make atmospheric correction
using ATCOR-2, followed by an automated classification applied for each image.
Finally, the mosaicing-process was applied for the produced thematic maps that
resulted from classifying each image. This mosaicing-process was helpful, in that
it made it easier to find the final statistical results for the whole study area, and to
compare the area with other results from separate data and dates.
5.2.4 Data Fusion
Image fusion is the process of fusing the lower multi-spectral spatial resolution
with the higher panchromatic spatial resolution, to generate a higher multi-spectral
resolution data set, which has the advantages of both: the high spatial resolution of
the panchromatic image; and the higher spectral resolution of the multi-spectral
image. It is one of the spatial enhancement techniques which are able to use the
corresponding information that obtained from different imagery about the same
terrain features in an effective way (Liu and Mason 2009 ).
Fusing panchromatic- and multispectral- data includes two general steps: (1)
the geometrically registration the low-resolution multispectral imagery to the high-
resolution panchromatic imagery (see Sect. 5.2.1 ); and (2) merging the informa-
tion contents, spatial and spectral, to produce one data set that have the best
characteristics of the two input data sets. Examples of image fusion techniques are:
IHS (Intensity-Hue-Saturation); PCS (Principal Component Substitution); HPF
(High-Pass Filter); RVS (Regression Variable Substitution); and SVR (Synthetic
Variable Ratio). In this study, the Gram Schmidt Spectral Sharpening Algorithm
was used.