[Previous] | [Session 55] | [Next]
J.J. Sudol, J.W. Harvey, C.G. Toner (National Solar Observatory, GONG Project)
Knowledge of the distortion in an image is important in a number of applications. This is especially true in helioseismology where errors in either time or position degrade inferences about internal solar structure and dynamics. Previously (BAAS 34(2), 732, 2002), we proposed a method for measuring the distortion in the images obtained with the Global Oscillation Network Group solar telescopes. After running a number of simulations, we have made some critical modifications to our method.
Our attempts to solve simultaneously for the image center and the distortion have proven to be unstable. We now solve first for the image center and then for the distortion. Methods that involve a large number of free parameters are imprecise because the solution space tends to be shallow and littered with local minima. The large number of calibration images and image points required to improve on this situation is impractical. In an effort to reduce the number of free parameters in our method, our error metric is now based on the differences in the distances between pairs of image points across all of the calibration images obtained.
In addition to these modifications, we now register the calibration images before and after correcting the images for distortion. The scatter in the image point positions in the first case sets an upper limit on the distortion and in the second case sets an upper limit on the extent to which corrections for the distortion can be made. In simulations, we have been able to correct for distortion to ~0.1 pixel, an order of magnitude above our desired correction of ~0.01 pixel, which is comparable to the error in determining the positions of the image points.
The author(s) of this abstract have provided an email address for comments about the abstract: jsudol@noao.edu
[Previous] | [Session 55] | [Next]
Bulletin of the American Astronomical Society, 34, #4
© 2002. The American Astronomical Soceity.