improve the contrast of the picture. The "histogram
equalization" method is commonly used for this purpose. This
transformation works with separate color components. Having a
normalized histogram h(x), such that
f
>*4
h(x) = 1
for one of the components, the transformed level H(x) for
primary level x will be given by the expression
x
H(x) = X. h(i).255
1*4
This transformation gives excellent results and can be
realized on the computer without problems.
Refining of the pixel size
The pixel size of TM data (30 x 30 m on the ground) is too
rough for 1 : 50 000) photomaps (1 pixel 0.6 mm). A lot of
different methods exists for refining of the pixel size. The
best results were obtained with our method based on the
Taylor's development.
Digital image may be understood as a function f(x,y) of
coordinates x,y. This function can be approximated with a new
function f*(x,y) using the grey levels of neighbouring pixels.
In case of using the Taylor's development, the expression for
the functiom f*(x+h1,y+h 2 ) will be
f *(x+hï ,y+h 2 ) = f(x,y) + —-( hijr- + h 2~fç) * f (x, y) +
+ T! (hl £ + *2fyz.f(x.y) 4- £(hl£ + h 2 ^)3.f(x,y)
.+ ... + + h 2 ) n * f ( x, y )
+
The partial derivatives are calculated as
J^f(x,y) = f(x+1,y) - f(x-1,y)
^f(x,y) = f(x,y+1) - f(x,y-1)
One pixel from the original data set will be substituted by
sexteen new ones with coordinates illustrated in the
following scheme
x, y are coordinates of the original pixel.
Color composite and its superimposition into the map
Before digital processing of TM data (according to the
226