Module API

OIS is a package to perform optimal image subtraction on astronomical images. It offers different methods to subtract images:

  • Modulated multi-Gaussian kernel
  • Delta basis kernel
  • Adaptive Delta Basis kernel

Each method can (optionally) simultaneously fit and remove common background.

Usage:

>>> from ois import optimal_system
>>> difference, optimalImage, optimalKernel, background =
                                optimal_system(image, referenceImage)

(c) Martin Beroiz email: <martinberoiz@gmail.com> University of Texas at San Antonio

exception ois.EvenSideKernelError
ois.convolve2d_adaptive(image, kernel, poly_degree)

Convolve image with the adaptive kernel of poly_degree degree.

ois.eval_adpative_kernel(kernel, x, y)

Return the adaptive kernel at position (x, y) = (col, row).

ois.optimal_system(image, refimage, kernelshape=(11, 11), bkgdegree=None, method='Bramich', gridshape=None, **kwargs)

Do Optimal Image Subtraction and return optimal image, kernel and background.

This is an implementation of a few Optimal Image Subtraction algorithms. They all (optionally) simultaneously fit a background.

Parameters:
  • gridshape – A tuple containing the number of vertical and horizontal divisions of a grid. Subtraction will be performed on each grid element. None is equivalent to a (1, 1) grid (no grid).
  • kernelshape – Shape of the kernel to use. Must be of odd size.
  • bkgdegree – Degree of the polynomial to fit the background. To turn off background fitting set this to None.
  • method –

    One of the following strings.

    • "Bramich": A Delta basis for the kernel (all pixels fit independently)
    • "AdaptiveBramich": Same as Bramich, but with a polynomial variation across the image. It needs the parameter poly_degree, which is the polynomial degree of the variation.
    • "Alard-Lupton": A modulated multi-Gaussian kernel. It needs the gausslist keyword.
  • poly_degree – Needed only for AdaptiveBramich. It is the degree of the polynomial for the kernel spatial variation.
  • gausslist –

    Needed only for Alard-Lupton. A list of dictionaries with info for the modulated multi-Gaussian. Dictionary keys are:

    • center: a (row, column) tuple for the center of the Gaussian. Default: kernel center.
    • modPolyDeg: the degree of the modulating polynomial. Default: 2
    • sx: sigma in x direction. Default: 2.
    • sy: sigma in y direction. Deafult: 2.

    All keys are optional. Example:

    gausslist=[{center: (5, 5), sx: 2., sy: 2., modPolyDeg: 3},
           {sx: 1.0, sy: 2.5, modPolyDeg: 1},
           {sx: 3.0, sy: 1.0},]
Returns:

difference, optimal_image, kernel, background
Raises:

EvenSideKernelError – If any dimension of kernelshape is even.