Photutils Python Package

Photutils is a Python library designed for detecting and performing photometry of astronomical sources. It is a coordinated package of Astropy and integrates with the broader Astropy ecosystem for astronomical image analysis.

Key capabilities include:

• Background Estimation: Tools for estimating the background level in images.
• Source Detection & Extraction: Star finding, source detection, and extraction.
• Photometry: Support for both Aperture Photometry and PSF (Point Spread Function) Photometry.
• Image Analysis: Image segmentation, centroid calculation, radial profiles, and elliptical isophote fitting.

Photutils is a Python package providing a wide range of tools for astronomical photometry and source analysis. Its primary capabilities include:

• Photometry: Performing aperture photometry and PSF-fitting photometry.
• Source Detection & Extraction: Detecting and extracting both point-like sources (e.g., stars) and extended sources (using image segmentation).
• Background Estimation: Estimating the background and background RMS in astronomical images.
• Source Characterization: Centroiding sources, estimating morphological parameters, and creating radial profiles or curves of growth.
• PSF Modeling: Building an effective Point Spread Function (ePSF) and matching PSF kernels.
• Galaxy Analysis: Fitting elliptical isophotes to galaxies.
• Image Simulation: Creating simulated astronomical images and estimating limiting depths.

The photutils.background module provides tools for estimating and modeling the background in astronomical images. This is a prerequisite step for many photometry tasks to ensure that source flux measurements are not biased by the underlying sky level.

For detailed workflows on how to select and apply specific background estimation methods, refer to the full User Guide.

The photutils.isophote module provides tools for elliptical isophote analysis. This is typically used to characterize the shape, orientation, and intensity profiles of astronomical objects by fitting ellipses to constant-intensity contours (isophotes).

For detailed workflows, including how to set up fitting parameters and interpret results, refer to the full User Guide.

A SegmentationImage is the output of detection and deblending processes. It is an object with the same shape as the input data where:

• Each detected source is assigned a unique positive integer label.
• Background pixels are assigned a value of 0.
• The segments represent the isophotal footprints of the detected sources.

An effective PSF (ePSF) is an empirical model that describes the fraction of a star's light that lands in a particular pixel. Unlike a standard instrumental PSF, an ePSF can be oversampled relative to the detector pixels.

Oversampling is critical when the PSF is undersampled by the detector (e.g., a FWHM of only 1-2 pixels). Because stars land at fractional pixel positions, an oversampled ePSF captures the 'pixel-phase variation,' allowing the PSF to be accurately interpolated to the exact position of any star.

When performing PSF matching with make_kernel or make_wiener_kernel, you can provide an optional window parameter to suppress high-frequency spatial noise artifacts in the Fourier ratio. A window function is a radially-symmetric 2D filter that equals 1.0 in the center and tapers to 0.0 at the edges, filtering out frequencies where the signal-to-noise ratio is low.

Both photutils.psf_matching.make_kernel and photutils.psf_matching.make_wiener_kernel accept a window argument. Note that windows are generally less critical when using make_wiener_kernel because its regularization already suppresses high-frequency noise.

# Example usage pattern (conceptual)
kernel = make_kernel(image1, image2, window=TukeyWindow(alpha=0.5))

Errors in photutils.isophote are calculated using two different approaches:

1. Direct Parameters: Most parameters computed directly at each isophote use standard error propagation.
2. Geometrical Parameters: Errors for the ellipse center, ellipticity, and position angle are estimated empirically. They are computed by "projecting" the harmonic amplitude errors (from the least-squares covariance matrix) using the same formulae used for parameter updates.