International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
3. DATA AND METHODOLOGY
3.1 Data
IRS 1 D Pan and LISS III images acquired on June 5, 1996 and
ERS image acquired on June 13, 1996 were used for the aim of
the study. 1/5000 scaled topographic maps and aerial
photographs were used for rectification process. Also, these
data were used to select training sites and perform accuracy
assessment during classification procedure.
3.2 Pre-processing
Digital image processing techniques were performed to extract
land cover categories from fused images in different phases.
[mage pre-processing was carried out including image
enhancement, and geometric correction. ERS radar image was
filtered using (3*3) mean and (5*5) median filter in order to
supress speckle effect and improve interpretation capabilities.
Geometric correction of all three images were performed using
first order polynomial equations. Image to map and image to
image registrations were applied to images in order to prepare
them for an accurate fusion application. IRS 1 D Pan image was
transformed into UTM coordinate system by means of GCPs
obtained by 1/5000 scaled standard topographic maps. LISS III
and ERS data were transformed to the same coordinate system
by using rectified IRS 1 D image. All images were resampled
to the same pixel size using nearest neighbourhood algorithm.
3.3 Combining high- and low resolution image data
After having transformed the dataset into the same coordinate
system, the images were fused to produce images which have
better spatial and spectral characteristics. The information
content of the resulting image may adversely changed by using
fusion techniques.
3.3.1 Brovey Transform
The Brovey Transform was developed to visually increase
contrast in the low and high ends of an image’s histogram.
Consequently, the Brovey Transform should not be used if
preserving the original scene radiometry is important. However,
it is good for producing RGB images with a higher degree of
contrast in the low and high ends of the image histogram and
for producing visually appealing images (Erdas Field Guide
1999). The Brovey transform is a simple method to merge data
from different sensors (Zhou, 1998). The formulae used are
shown in equations (1), (2), and (3):
RED = band 5/(band 2+band 4+band 5) * Pan (1)
GREEN = band 4/(band 2+band 4+band 5) * Pan (2)
BLUE = band 2/(band 2+band 4+band 5)* Pan (3)
3.3.2 Multiplicative
The multiplicative algorithm is derived by using the four
possible arithmetic methods to incorporate an intensity image
into a chromatic image (addition, subtraction, division, and
multiplication), only multiplication is unlikely to distort the
color. Equation of the algorithm is as follow;
(DNMSn)*(DNPAN) = DNnew*MSn (4)
576
DNMSn- Digital number of pixel belongs to n-th multi spectral
band
DNPAN- Digital number of corresponding pixel belongs to
panchromatic band
DNnew MSn- New digital number of corresponding pixel
(Erdas, 1999).
3.3.3 Principal Component Analysis (PCA)
The purpose of PCA is to compress all of the information
contained in an original n-band data set info fewer then n “new
bands" or components. These components are computed by
linear combinations of the original images. None of the
component is linearly correlated with other because these n
components are orthogonal. The total variance of original
images is mapped onto new components. The first principal
component (PCI) has the greatest percentage of the total
variance and succeeding components (PC2, PC3,.., PCn) each
contains a decreasing percentage of the total variance (Lillesand
and Kiefer, 2000; Lucian Wald, 2002).
3.3.4 Intensity Hue Saturation Transformation
The general THS procedure uses three bands of a lower spatial
resolution dataset and transforms these data to IHS space. This
numerical procedure was developed to convert a three-band
RGB (red-green-blue) display into its fundamental
physiological (IHS) elements of human color perception
(Grasso 1993). The IHS color coordinate system is based on a
hypothetical color sphere. The vertical axis represents intensity,
which ranges from 0 (black) to 255 (white). The circumference
of the sphere represents hue, which is the dominant wavelength
of color. Hue ranges from 0 at the midpoint of red tones through
green, blue and back to 255, adjacent to 0. Saturation represents
the purity of the color and ranges from 0 at the center of the
color sphere to 255 at the circumference (Jensen, 1996). The
IHS values can be derived from the RGB values through
transformation equations. In order to apply this technique for
the enhancement of spatial resolution, a panchromatic higher
resolution channel replaces the intensity component of a lower
resolution multi spectral dataset.
3.4 Comparison of the spectral and spatial effects of the
merged IRS 1 D Pan- LISS III and ERS — LISS III images
3.4.1 Visual comparison
Visual image interpretation was performed for comparison of
the merged images. Visual interpretation analysis showed that
merged images of IRS 1 D pan and LISS III have better spatial
and spectral details then original IRS 1 D Pan and LISS [II
images. Especially, buildings, roads and crossroads were
identified easily from merged images compared to original
LISS III image. Visual interpretation of ERS image was
improved by merging ERS and LISS TIT images. Figure 3 shows
all images which were merged by using different fusion
techniques. Same section of the fused optic images were
examined and interpratation results showed that Multiplicative
method caused smoothing
Intern