Image Subtraction Noise Reduction Using Point Spread Function Cross-correlation

TL;DR

This paper introduces a cross-correlation technique to enhance noise reduction in image subtraction for astronomy, improving the detection of transient objects by refining PSF matching.

Contribution

It proposes a novel cross-correlation method to reduce noise in delta function-based convolution kernels for better image subtraction quality.

Findings

Significantly improved signal-to-noise ratio in image subtraction results

Enhanced detection capability for transient astronomical objects

Effective handling of spatial PSF variations

Abstract

Image subtraction in astronomy is a tool for transient object discovery and characterization, particularly useful in wide fields, and is well suited for moving or photometrically varying objects such as asteroids, extra-solar planets and supernovae. A convolution technique is used to match point spread functions (PSFs) between images of the same field taken at different times prior to pixel-by-pixel subtraction. Particularly suitable for large-scale images is a spatially-varying kernel, where the convolution is allowed to adapt to PSF changes as a function of position within the images. The most versatile basis for fitting the spatially-varying kernel is the Dirac delta function. However, the convolution kernel based on the delta function does not discriminate between pixel scale noise variations and the intended stellar point spread function signals. The situation can frequently lead to reduced signal to noise ratios for variable objects detectable in the resulting subtraction. This work presents a cross-correlation method for reducing noise effects on the delta function derived convolution kernels, thus yielding significantly improved signal to noise in the resulting subtraction.