imProc.sub - EranOfek/AstroPack GitHub Wiki

imProc.sub - Image subtraction tools

Background

Typically transient detection is done using image subtraction (or image differencing). Since typical astronomical images does not have the same PSF, direct subtraction is not effective.

imProc.sub contains high-level functions (that works on AstroImage objects) related to image subtraction and transient detection. The core functionality that is being used by this package is implemented in the imUtil.properSub package.

The functions implement the proper image subtraction algorithm (Zackay, Ofek, and Gal-Yam 2016; hereafter ZOGY), and the translient algorithms (In prep.).

ZOGY formula is a simple hypothesis testing between the null hypothesis that there is no transient in some position q, and the alternative hypothesis that there is a transient in this position. Similarly, the translient statistics were designed for testing the motion hypothesis.

Terminology

  • S - The ZOGY image subtraction statistics (usually in units of sigmas). In each pixel in S contains the significance that a transient in this position is real. Since S is the result of a matched filter operation, it is correlated. A local maximum which value is larger than 5, or smaller than -5, can be regarded as a 5-sigma transient candidate. However, S assumes no astrometric errors and that the sources are in the background-noise-dominated regime.

  • D - The ZOGY proper subtraction image - Here we define a proper image, as an image which pixels are not correlated. One can get S, by cross-correlating D with its PSF (Pd).

  • Pd - The PSF of the D image.

  • Scorr - Corrected S image. Since S does not include source noise and astrometruc noise, Scorr is a version of S in which the source noise, or astrometric noise (or both) was propagated into its noise. Therefore, source noise, or astrometric noise artifacts should not appear in this image.

  • Z2 - The Z-square translient statistics is the translient score for motion of the source in a non-particular direction. Z2 values distribute like a $\chi^{2}$-distribution with two degrees of freedom.

  • Z2sig - A version of the Z2 image that was converted to units of Gaussian significance.

Functions

Example