Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
inconsistent results after flipping an image
01-25-2016, 03:19
Post: #1
inconsistent results after flipping an image
Hello,

when using SExtractor v2.19.5, I am finding inconsistent results when comparing an image to a flipped version of itself. [By 'flipped', I mean the x-pixels have been mirrored, e.g., (1,117) -> (2048,117) and (2048,112) -> (1,112).]

Despite the only changes having occurred in the x-direction, the output for Y_IMAGE (or Y_IMAGE_DBL) can be different by more than 0.1 pixels, and radius measurements (FLUX_RADIUS, PETRO_RADIUS, etc.) can be different by more than 0.5 units (even when A_IMAGE and ELONGATION are the same). Some magnitudes (e.g., MAG_APER, MAG_PETRO) can be different by more than 0.5mag.

I've also come across circumstances where an object was split into 2 detections in one orientation but not the other, despite running SExtractor with identical configurations.

These issues seem to occur on both linux (CentOS 6.7 and 7.1.1503) and OS X (10.7.5). They also occur with both the stable (2015-06-29) release and the latest development version (r332, 2016-01-25). I've tried multiple ways of flipping the images and have also stripped the image headers bare to avoid any associated problems, but the differences persist.

Has anyone seen that before? Any ideas on why that might be, or how we can avoid such inconsistencies?

Thanks,
Chris
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)