A note on artificial deformation in object shapes due to the pixelization

TL;DR

This paper investigates how pixelization effects, especially during resampling with rotation, induce artificial shape deformations in astronomical objects, potentially affecting weak lensing measurements.

Contribution

It provides a detailed analysis of pixelization-induced shape distortions and their dependence on object size, resampling schemes, and pixel grid size, highlighting their impact on weak lensing.

Findings

Artificial ellipticities can be significant for small objects (FWHM < 2 pixels).

Shape deformation depends on interpolation scheme and pixel grid size.

Proper interpolation can suppress pixelization effects below 10^-2.

Abstract

We qualitatively examine properties of artificial deformation in shapes of objects (galaxies and stars) induced by the pixelization effects (also called as the aliasing effects) using toy mock simulation images. We pay a special attention to the the second pixelization because it might be a potential source of a systematic noise in a weak lensing analysis. In particular, it is found that resampling with rotation induces artificial ellipticities in object shapes having a periodic concentric-circle-shaped pattern. Our major findings are as follows. (1) Root-mean-square (RMS) of artificial ellipticities in object shapes induced by the first pixelization effect can be as large as RMS > 10^-2 if a characteristic size of objects (e.g., the FWHM) is smaller than twice of the pixel size. While for larger objects, it quickly becomes very small (RMS < 10^-5). (2) The amplitude of the shape deformation induced by the second pixelization effect depends on the object size. It also depends strongly on an interpolation scheme adopted to carry out resampling and on the grid size of the output pixels. The RMS of ellipticities in object shapes induced by the second pixelization effect can be suppressed to well below 10^{-2} if one adopts a proper interpolation scheme (implemented in popular image processing softwares). We also discuss an impact of the pixelization effects on a weak lensing analysis.