wo
ire
13
er
id
In the spring term (duration 13 weeks) the course SATK2 is
offered. This is more specialized, and is aimed primarily at
students of mapping sciences. In this course ERDAS is used in
all the four following exercises:
Satellite image geometry.
Determination of terrain elevation from a single
oblique SPOT scene.
Special image-processing procedures.
Colour-hardcopy production, Unsupervised
classification (clustering) and Field-control
(collection of ground-truth).
All the exercises have in common that they are based on
subsets taken from different types of satellite-scenes in digital
format. After processing of the initial data-sets by the various
ERDAS-functions according to the guidelines given in the
exercise text, the result will be one or more new image files. It
will often be important to compare two or more images. This
is obtained by displaying the images on the RGB-screen
simultanously. In addition to a written description of the work
carried out during the exercise, the students also deliver a copy
of the "Result Screen-Images" on a diskette. PIC is useful for
saving the "Result Screen-Image" as an image file.
The descriptions of the exercises given below are accompanied
by illustrations of such "Result Screen-Images”. The
illustrations in this paper are produced as B/W hard-copies on
the Tektronix plotter at reduced scale. The quality of these
illustrations is of course much lower here than on the screen,
but it is hoped that they will contribute to a better
understanding of the use of ERDAS in the exercises.
SATKI. Exercise 1. Digital image processing.
In this exercise the students get in touch with ERDAS for the
first time. The aim of the exercise is to familiarize the students
with ERDAS and with basic concepts and procedures within the
field of digital image processing. The main emphasis in this
exercise is on image enhancement methods.
The initial satellite image is a subset of a panchromatic SPOT
scene covering the area where the University is located. The
subset measures 768 lines by 768 pixels. The first step is to
show the image on the screen using READ. (The file
containing the statistics for this initial image has already been
generated). Using CURBOX and SUBSET, a secondary subset
of 200 x 200 pixels is produced. A -.STA-file is generated by
BSTATS, and the functions DHISTOEQ, STRETCH and
DCONVLV are then used to produce enhanced derivatives of
the original. These images are then compared in the "Result
Screen-Image" as shown in Figure 1a. CURSES is used for the
comparison of corresponding pixel-values in the differently
stretched images.
Figure la. Various types of contrast stretch enhancement of a
panchromatic SPOT subscene.
In the second part of this exercise, the initial image is a subset
of a Landsat TM scene. The data-set contains 6 bands, as the
thermal band with the coarser resolution is removed. For further
processing, as in the first part, a secondary subset is generated.
In addition to apply the skill obtained during the first part of
the exercise, the principal components transform PRINCE is
used. The "Result Screen-Image" is shown in Figure 1b. Text
is written by ANNOTAT.
Figure 1b. False Colour Composite and single band display of
selected original and principal components Landsat
TM data.