flatmaker2 - program for making flats

flatmaker2 - program description

1. General description

It is program for creating flat field frame from series of sky images. It requires a lot of memory , as all frames are read to memory however SMOK
with 4GB of ram will allow to calculate median of 4000 / 8.5 ~= 470 frames , so 400 should make no problem ! This should give a perfect flat !
There are still some things to be improved here - red ones.

Main steps :

read all frames into memory
for each frame calculate baground map , function CCDMatrix::createBackgroundLuminosityMap , which creates bacground map
of 30x30 pixel squares, generally it is calulated as median value of first 50 pixels in map field (30x30 ) this is quite strange , maybe
all 90 pixels should be used instead , or somehow randomly but not first 50 pixels !!! , only pixels with
VALUE < MAX_LIMIT = third_value( 15 map pixels ) + prog
are used in this median calculation.

frame values are divided by average value of bacground on frame ( in fact it is done in ReadLine function - when reading next row )
loop over y {
    ReadLine(y)
    loop over x {
      FLAT(x,y) = FLAT(x,y,METHOD);
    }
}
where METHOD is defined as paratemeter -type 0/1/2
METHODS :
0/ Bogumils's Method :
   - takes values from all frames of pixel (x,y) to table lum[] , sorts it and finds limit :
     lum_hi = (lum[lum_count/4]+lum_add) * lum_mul;
   if lum_hi>2.5 (or 4.0) error is claimed
   - use pixels <=lum_hi limit and find median value which is used as final FLAT_PIXEL(x,y) value
   - claim error if : pixel_count<8 , median<0.35 or median>1.6
   - return median
1/ - Marcin's Sokolowski method :
    This method is very simple , values of pixel (x,y) in table lum[] are sorted and median value
    is found , but only of first 75% of pixels ( lower part ) to ignore higher values from stars.
    So this is :
          median := lum[ (0.75*lum_count)/2 ];
2/ fit of gauss to each pixel - very time consuming method :
    lum[] table is filled ( not sorted) and distribution of pixels is created, gauss is fitted.
    and final FLAT_PIXEL(x,y) = mean of fitted gauss
    This is not very well working method, maybe much more frames are needed to have really nice gauss !
     NOTE : this option is not working very well , it requires a lot of frames this means a lot of memory, I tested it with 200 frames
      and result was much worse then option type=0

Testowalem ostatnio na smok(heplx45) /data2/results/tmp/FLAT - calkiem przyzwoicie zadzialala

2. Program options

Program is called with name of file with list of fits files from which flat should be produced and name of output FLAT field name, also name of
dark fits file should be provided, example :

flatmaker2 k2a_list flat_k2a_85mm.fit -dark=dark1.fitc

Example of flat field creation procedure is described here.

Option Name	Description	DEFAULT
-lum_add	Used in limit lum_hi determination value added to value in 1/4 of sorted table of pixel (x,y) values from each frame	0.1
-lum_mul	Used in limit lum_hi determination, multiplier of (lum[lum_count/4]+lum_add)	1.2
-lum_minhi	Used in limit lum_hi determination, if calculated lum_hi < lum_minhi then lum_minhi is used	0.5
-prog	Used in bacground calculation, only pixels below this limit are used in median bacground calculation in map field : max_limit = square_third_value + prog_value pixels above such max_limit are ignored	125
-border	Size of border to be skiped from making flat	30 - must be changed to make flat of all pixels without any borders !
-frame	Definition of borders in image as : -frame=(left,up,right,bottom)	24,8,8,8
-dark	path to dark frame file	no dark
-type	type of flat calculation 0 - Bogumil's, 1 - msok, 2 - fit of gauss to each pixel	0 - Bogumil's method
-verb	verbose mode	0