Opened 16 years ago
Closed 15 years ago
#1381 closed defect (fixed)
dvoImagesAtCoords can find coordinates in 2 chips
| Reported by: | bills | Owned by: | bills |
|---|---|---|---|
| Priority: | normal | Milestone: | |
| Component: | DVO | Version: | |
| Severity: | normal | Keywords: | |
| Cc: |
Description
dvoImagesAtCoords takes an smf file and a list of ra and dec
values and determines which chip (if any) contain the coordinates.
It does this by transforming the ra and dec to chip coordinates and then comparing that to the extent of the chip.
In certain cases the astrometric solution does not extrapolate well for coordinates far from the chip. Here is an example
dvoImagesAtCoords -astrom /data/ipp016.0/nebulous/7b/cb/267906035.gpc1:ThreePi.nt:2010:03:26:o5281g0257o.151659:SR_o5281g0257o.151659.cm.63547.smf 187.315827 34.006668
1 187.315827 34.006668 XY10
1 187.315827 34.006668 XY65
The correct answer is XY65.
Gene suggests that we change the method that we use to compute the overlap.
Transform RA DEC to "focal plane coordinates" using the fpa level transform. Then use the layout of the chips to on the focal plane
to at least get close to the chip. Then use the existing algorithm
to confirm that the coordinates are in the area covered by the chip.
I have fixed this problem.
The check for overlap in image coordinates is done as before, but when a match is found I transform the ra/dec and the extents of the image to LM (focal plane) coordinates.
If it is an actual overlap, the point will be within bounds. If the overlap is bogus, this check will catch it.
Confirming after the match is found greatly reduces the amount of computation in the usual case (no overlap).