Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.izmccd.puldb.ru/izmccd4e.htm
Дата изменения: Mon Sep 17 19:47:52 2007 Дата индексирования: Mon Oct 1 19:27:50 2012 Кодировка: Поисковые слова: ngc 6781 |
Control and improvement of accuracy of results.
The process of improvement of accuracy suppose
a) a choice of the best mode of measurements of stars and objects ("Properties/Measurement"),
b) a choice of reference stars.
In errors of determination of equatorial coordinates of objects, the important
part plays aberration of a projection of a telescope or generalized distortion.
Because of similar defects a shifts of coordinates can reach into several
tenth second of an arch. If you will use instead of the command "Calculate
coordinates" the command "Calculate coordinates L.", you can
reduce this effect partly. The local astrometric solution will be used in
the latter case. Thus reference stars will be located near the object. Minuses
of this method are
1) necessity of presence on an image of a plenty of stars (we recommend to
apply this mode, if on the frame of reference stars is more 100)
2) the reduction of number of reference stars results in deterioration of
accuracy of coordinates in a random quotient.
More correctly, to apply correction for this effect beforehand. This correction will depend on coordinates in system of an image only. The program takes the corrections from file "corrfld.bin". This file is developed separately for each telescope, and should be updated both after the expiration of the certain interval of time, and after anyone adjustment of a telescope. It is necessary to make research of properties of your telescope for creation of the specific file. Address to the author of the program - .
The window is printed on the screen at definition of equatorial coordinates:
Here
Sx is accuracy of measurements X-direction (here axis X corresponds
to a direction to east).
Sx is accuracy of measurements Y.
Mx, My are scales in X and Y.
Gamma is angle of oblique frame.
Theta is angle between a direction to east and axis X of a matrix.
If a Sx and Sy are less, the accuracy of measurements is better. Sx and Sy
depend from residuals at all reference stars (reference stars are marked by
blue color).
If you want to see residuals:
1) Give the command "View/Information", or click .
2) Click on any reference star. Residuals are designated dsx and dsy.
Other opportunity is:
Select the menu "Calculations/Residuals", or .
In this window N is number of a star; X, Y are measured coordinates; m1, m2
are stars magnitudes. If you will click on ; you exclude from calculations the corresponding
reference star.
You may exclude in size of residual, if residual will be more than in 2 times,
than value S.
Also it is possible to exclude either darkest or brightest stars; if these
stars have residuals more than average. After you will press the button "OK"
again there will be the calculation of coordinates.
At measurements of an images of stars a residuals of other of kind are calculated.
These residuals are the image of the star a minus function of a profile for
each pixel. You can supervise quality of measurements of the star by a control
of these residuals. See section "Basic algorithms
of program". In the program two ways are realized.
1) Select the menu "View/Profile's residuals". The images of stars (already measured) will look as group of pixels of red and blue color. If a residual on this pixel is more than zero; the pixel has red color differently blue. The best distribution of residuals is a casual distribution:
.
The acceptable distribution is circular distribution –
To convert circular in casual it is possible for one of ways
a) Reduce value of "Property/Measurement/Aperture", but the casual error of measurements will be worsened.
b) Change value "Property/Measurement/Exponent of power". This size adjusts the form of fall of brightness in function of a profile.
c) Set check box "Property/Measurement/ Exponent of power is variable".
d) Change here a profile of Lorenz to a profile of Moffat or back to.
If the structure of the red-blue image has the allocated direction, it is the most bad variant:
.
Thus the measured coordinates will have shift concerning a true placement of a star in this direction.
2) Open the menu "View", select "Profile". The window will appear:
In this window the correlation between intensity and distance of an element up to the centre of the image is outputted. A function of a profile looks smooth curve in such picture, and an observational data give disorder of points concerning some average course (everyone pixel of the image of a star give one point). The data are outputted at the moment of measurement. If points surround the curve, it is ideal variant. Differently set of possible actions is same as in item 1.