Iterative Beam Removal

When developing an algorithm for detecting streaked reference stars, we tried multiple methods - Maxima Detection and Iterative Beam Removal - before testing the methods implemented in Astreaks. The workflow of Iterative Beam Removal is discussed below in detail, followed by a brief discussion of the shortcomings of this method. The code for its implementation can be found at https://github.com/krittisharma/iterative-beam-subtraction.git.

Workflow of Iterative Beam Removal

The major steps in the workflow of Iterative Beam Subtraction include rotating the original image and generation of Streak Spread Function (SSF) Model, cross-correlation of the original image with SSF model, iteratively searching for maxima and removing the beams until a threshold is reached, injecting flux scaled detected sources in a synthetic image and solving it for WCS.

Calibrated Data

The data is calibrated by bias correction and flat fielding. The image is then rotated to align the stellar streaks with North. The rotation angle can be computed using the telescope position angle and velocity position angle. This is done primarily to reduce the size of the kernel (PSF Model) to save computation time of convolution to generate the SSF model and cross-correlation thereafter.

Cross-Correlation

Next, we cross-correlate the original rotated image with the Streak Spread Function model, which is computed using the same methodology, as that of Astreaks workflow. This is done with a two-fold motivation:

Enhance signals from weak sources
Smooth the image for the detection process

1st Iteration

We search for 1st maxima in the cross-correlated image and remove 90% intensity of the corresponding streak. After removing the beam, we cross-correlate it again with the SSF Model.

After 30 Iterations

We observe that there is not more significant signal left for detection in the original image. Even in the cross-correlated image, everything is just a blur. Hence, we stop iterating at this point and move on to the further steps.

Synthetic Image Generation

Flux-scaled synthetic sources are injected into a synthetic image and the background is added. We then solve this synthetic image for astrometry and use WCS of the solved image for the astrometric measurements of the minor planet at the mid-time of the exposure.

Page updated

Google Sites

Report abuse