Description
Roughly circumscribes the planet by estimating points on the bright limb and then fitting a circle. The limb estimation uses the technique from 'findlimb.py'. A circle is then fit to these points using the technique from 'fitcircle.py'. The resulting coordinates (cx, cy, radius), fit error (RMSD in radius), and number of points used are printed.
Usage
roughalign.py fits-file [thresh dir interval0 interval1 min_count], or roughalign.py folder [min-fits-file max-fits-file thresh dir interval0 interval1 min_count]
This program can operate on a single file or on a sequence of files. On a single file, it estimates a set of points on the bright limb, and then fits a circle to those points, printing the resulting coordinates and error, and plotting the points and circle over the input image. On a sequence of files, it outputs a list of coordinates that can be redirected to a text file. Since the limb estimation relies on negative row values in the image (in the saturated crescent), this program is best used on raw images.
'roughalign.py' attempts to find the limb in every 8th row of the image. Using more rows doesn't improve the result, but runs slower. 100 iterations of the circle fitting algorithm are used, whereas the error converges in about 60 iterations. All of these settings can be changed in the code.
The program contains a function which can be called from other programs:
roughAlign(im, thresh=0.75, dir=0, interval=(50, 5), minCount=5, printout=False, returnPts=False) 'thresh' is the fraction (0.0 to 1.0) of the row max that a peak must be. 'dir' is the orientation of the bright limb. 0 = sun to the right, 1 = sun to the left. 'interval0' is the number of pixels to the left of each peak. 'interval1' is the number of pixels to the right of each peak. 'min_count' is the minimum number of negative values which must be in the crescent side of the peak. 'returnPts' is a flag which returns the limb points as well as the resulting circle. See the 'findlimb.py' page for details.
Examples
This program works reasonably well on most raw images:
> roughalign.py 072407/raw/im0215.a.fits Circle: cx=296.959917 cy=253.505009 rad=210.532756 err=1.921364 npts=38
It can even accomodate some images where the bright limb is partially offscreen:
> roughalign.py 090807/raw/im1850.a.fits 0.75 1 5 50 5 Circle: cx=155.056316 cy=267.498732 rad=214.335478 err=0.371155 npts=9
> roughalign.py 090807/raw/im1979.a.fits 0.75 1 5 50 5 Circle: cx=385.863713 cy=233.859617 rad=212.945557 err=0.889931 npts=44
For sun on the right, the default parameters should suffice (thresh=0.75, dir=0, interval0=50, interval1=5, min_count=5). For sun on the left, set dir=1 and reverse interval0 and interval1 (e.g. args='0.75 1 5 50 5'). This program should work on both cont-K and br-gm data. Play with it to see how far the bright limb can be out of the frame before there are not enough limb points for a good circle.
Multiple files:
> roughalign.py 090807/raw im1970.a.fits im1979.a.fits 0.75 1 5 50 5 090807/raw/im1970.a.fits 378.608389 233.739025 214.544781 1.808101 43 090807/raw/im1971.a.fits 377.506105 236.038080 212.303412 1.854545 45 090807/raw/im1972.a.fits 375.062214 233.379800 211.627523 1.903415 46 090807/raw/im1973.a.fits 380.392327 233.873925 214.502585 1.012246 40 090807/raw/im1974.a.fits 374.509289 234.231681 210.227520 0.933074 41 090807/raw/im1975.a.fits 380.821674 234.526641 215.306581 1.193662 45 090807/raw/im1976.a.fits 382.243533 233.505755 211.512986 1.072806 36 090807/raw/im1977.a.fits 381.191627 235.110605 212.602575 1.459239 43 090807/raw/im1978.a.fits 385.618460 231.851314 213.044879 1.183417 42 090807/raw/im1979.a.fits 385.863713 233.859617 212.945557 0.889931 44 > roughalign.py 072407/raw 072407/raw/im0001.a.fits 225.237488 282.057854 199.038452 0.480574 40 072407/raw/im0002.a.fits 224.735053 281.393230 198.938637 0.662971 44 072407/raw/im0003.a.fits 226.727053 281.324609 198.203339 0.569237 38 072407/raw/im0004.a.fits 227.656998 280.704849 198.443130 0.657364 41 072407/raw/im0005.a.fits 226.821066 281.268604 198.693074 0.499104 37 072407/raw/im0006.a.fits 226.667954 280.817715 198.395979 0.348853 36 072407/raw/im0007.a.fits 225.496243 281.141945 198.837367 0.450066 39 072407/raw/im0008.a.fits 223.831192 280.181402 199.399277 0.538154 41 072407/raw/im0009.a.fits 224.425934 281.170340 200.107891 0.428056 32 072407/raw/im0010.a.fits 227.008706 280.204848 198.405911 0.549547 42 072407/raw/im0011.a.fits 228.420511 281.106945 198.177213 0.575066 39 072407/raw/im0012.a.fits 229.346603 281.179160 197.424958 0.571816 36 etc...
Note the increase in planet radius from 7/07 to 9/07. Output is only printed for files where >=3 non-colinear points can be found. By looking at the radius field, you can tell when the program is getting a good limb:
> roughalign.py 090807/raw im0001.a.fits im0351.a.fits 0.75 1 5 50 5 090807/raw/im0092.a.fits -23.125000 512.312500 84.052244 1.942289 4 090807/raw/im0095.a.fits 60.283854 396.039062 68.203261 0.076719 3 090807/raw/im0097.a.fits 54.614583 385.687500 79.391036 0.396454 3 090807/raw/im0101.a.fits 48.927083 401.828125 55.817203 0.178176 3 090807/raw/im0103.a.fits 67.843750 361.906250 96.130753 0.614768 3 ... 090807/raw/im0139.a.fits 127.541667 294.767361 145.502579 1.196338 9 090807/raw/im0140.a.fits 171.681822 244.654761 215.063554 0.708980 9 090807/raw/im0141.a.fits 168.556152 247.537109 217.595490 0.920200 8 090807/raw/im0142.a.fits 107.102679 326.312500 119.383888 1.066484 7 090807/raw/im0143.a.fits 119.919643 308.343750 136.477122 0.658026 7 090807/raw/im0144.a.fits 139.687500 285.937500 165.706108 1.298860 8 090807/raw/im0145.a.fits 164.512817 246.350586 213.786221 0.393891 8 090807/raw/im0146.a.fits 168.696289 248.621094 218.414932 0.494656 8 090807/raw/im0147.a.fits 173.728223 245.920215 212.351439 2.045703 10 090807/raw/im0148.a.fits 179.991721 242.614839 215.202593 0.774199 11 090807/raw/im0149.a.fits 189.277464 249.379311 219.034582 1.190396 11 ...
©Sky Coyote 2007