CO Band Interleaved by Line
; Number of Plxeis O Band Inter-eaved by Pixel
i m n a Pixel Interval Distence [m]
_
| |® Image Date Attached to the Header
Line interval Distance [mj
: © image Date in Separate Files
Figure 1. Data conversion menu
3.2 Image Display
There are two image file types that can be displayed:
WinASEAN GIH and classified image WinASEAN CLS.
WinASEAN GIH file is displayed by using Image display and
classified image by Classified image display menu. Programs
automatically detects whether input file is of suitable format for
display or not. Another function for image display is Pseudo
Color Image Display that is actually level slicing. The original
GIH image is displayed in true color mode so number of slices
can be up to 255. User selects certain gray value range and
assigns it to arbitrary color.
Under the Image display menu is there another function
Image Display and Print. By this function user can display two
image types: full color image with extension .GIH and classified
image with extension .CLS and send them to system printer at
selected scale. This is not hard copy as often used in many
Windows application by PrintScreen or Alt+PrintScreen
function. In order to print a image at exact scale program
requires following input data:
- Ground resolution of image (usually in meters)
- Scale factor
- Technical parameters of the system printer (supported
by Windows)
After program was informed about the above data, actual
ground coverage of the image will be computed and
transformed into printer resolution unit (dots per inch). The size
of the selected area to be printed is input not in image but in
printer resolution dimension.
3.3 Image Encoding
Image encoding is a special technique used by
WinASEAN to improve calculation speed of classification,
image transform etc. After encoding, data in the WinASEAN
GIH format is converted to WinASEAN CPT format with
structure as follow :
- Image header 1024 bytes
- Encoded image data Number of lines * number of
columns * code size in bytes
- LUTI Look - up - table containing all
Unique Pixel vectors ( UPV )
- LUT2 UPV frequency
The code size used for encoded image data is of 2 or 3
bytes. UPV frequency is hold in table with type of 1, 2 or 3
bytes. During processing program works with look-up-table
LUTI and LUT2 insteed with whole of image and then by index
replacing technique, result image, for example, classified or
principal component will be created.
| Green Assigment. »
| Blue Assigment.. b
Figure 2. Image display menu
3.3 Arithmetic Operation
Inter-channel or inter-image operation are very useful for
image enhancement, feature extraction and multitemporal image
analysis. Addition, subtraction, multiplication, division and their
combination can be applied for many purposes, including noise
elimination. In order to carry out the most different arithmetic
operation either among bands of one image or among different
bands of different images, the algorithm is divided into 3 steps.
In the first step program prompts for input data, then a temporal
image in .GIH format is created. In the second step program
asks user to give actual formula which defines the arithmetic
operation to be considered. In fact the formula is not limited on
basic arithmetic operation only, but any mathematics expression
of length up to 60 characters long and consisted of FORTRAN
intrinsic functions can be supplied. The construction of the
formula must follow strictly FORTRAN syntax. Program checks
syntax of the input formula and if there is no error it proceeds to
next step when actual computation will be realized. In order to
prevent division by zero, all digital counts having values 0 is
assigned 1 before calculation.
3.4 Multispectral Classification
Multispectral classification is based on the Maximum
likelihood classification method and consists of the following
operation:
- Training area selection
- Training data statistics calculation
- Training data redisplay for checking
- Training data modification (if necessary)
After checking training data and statistics computation the
maximum likelihood classification can be carried out. Input files
are image file in CPT format and training data (statistics) file.
Output will be classified image in CLS format. WinASEAN 2.0
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B6. Vienna 1996
can
unk
3.5
mul
betv
