The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1152
The main weakness of ratio image is lost the reflection
intensity information of ground objects. In order to remedy the
deficiency, one origin band and two ratio images can be
combined into one color image[3].
In this paper, the method of combining two ratio images with
one original band is researched. Because this method reserve
not only absolute radiometric information which have missed
in ratio images, but also some terrain detail information which
can distinguish some objects on the ground. It is easier for
people to interpretation.
Figure.2 False Color Image
Figure 2 is a hybrid false color image . In the image ,the Red
band is the ratio of band five and band one ,and the Green band
is original band one ,and the Blue band is the ratio of band five
and band one . The image is combinated by two ratio images
and one original band together which show us a better effect
one . Not only can it show us some important things on the
ground, but also some details of terrain. The effect is better
than that of tradition method dealing with in specialty fields
such as mineral.
However, this method still has some disadvantages^].
Abnormal of tonality will cause some difference of color in the
image. In RGB color system, color is defined by the proportion
of tricolor. The change of each one of tricolor will cause that of
color characteristic. The color can hardly described by number
well and truly. And the levels of number can hardly expressed
by a certain color, too.
4. THE OPTIMUM INDEX FACTOR(OIF) METHOD
Because the combination of R,G,B band is really complex and
changeable, the final result of image is different. In some cases,
it is difficult to choose a appropriate combination scheme for
false color composite of ratio images or select a color to
describe the specific information.
How to select the most useful combination achieving the best
reinforcing effect? Nowadays two methods have been
adopted[5]. First is to select the most suitable bands which can
identify different geologic body. The other is to select the
bands depended on statistical characteristics of images.
4.1 the Formula of OIF Method
In this paper, through researching the effect of ratio false color
composite image of RS images, the method of OIF (The
Optimum Index Factor) is discussed. Chavez ec al. has been
put forward a formula to calculate OIF(The Optimum Index
Factor) by using three standard deviation and correlation
coefficient.
OIF = Y d SDi/jfCC]\
i=1 J=l (1)
In the formula (1), SDi means the standard deviation. |CCi|
means the absolute value of correlation coefficient. Standard
deviation represents the discrete degree. The more value it will
be ,the more radiation intensity difference we can see. And the
less of the |CCi| ,the less of repeated degree .
The bigger OIF value we have calculated, the more information
the color image contains. And the less redundancy information
it is. So it is easy to understand that the biggest OIF value
represents the prime combination.
4.2 the Disadvantages of OIF Method
In the paper, in order to find a suitable combination scheme, the
OIF method is used to evaluating them objectively. Although
the method of OIF value is effective on representing the display
effect and narrowing the possible wave bands combination, the
disadvantages are very obvious[6,7].
(1) The OIF value embodies that of image information entropy
of from one side and there is no necessary association between
the value of OIF and image quality evaluation.
(2) Some images with similar bands combination have similar
OIF value, unfortunately their display effects are dissimilar
obviously.
(3) The best image combination maybe not just the image of
the analyst wanted with special information.
5. RELATIONSHIP BETWEEN THE OIF VALUE AND
THE SCALES OF THE WINDOW
First two ratio image we talked above are filtered by sliding
window method with different scales. Then two new
characteristic matrix we have got are synthesized into a false
color image with a original band. Finally we calculate the
OIF value.
For the sake of discussion, a simple expression is provided.
R=band51 mean a new characteristic matrix which can be got
by filtered with different scales of window after band one
divided by band five . The scales of filter window are set in
different sizes. Such as ^ x 3 ^ 5x5^ 7 X 7 ^ 9x9
11x11 13x13 15x15 17x17 19x19^
21x21, 31x31^ 41x41 Based on the formula of OIF,
the OIF value is closely related not only the scale of sliding
window but also the choice of original band [8,9].
Research area of RS image is Tieshan mining area in Daye city ,
Hubei province. Because it is a big opencast mine area where
the research data is relatively comprehensive. And the
prospecting criteria alteration zones can be found easily
which will affect some RS bands a lot. So the OIF value is