Wavelet-based image decomposition method for NuSTAR stray light background studies

Andrey Mukhin,Roman Krivonos,Alexey Vikhlinin,Brian Grefenstette,Kristin Madsen,Daniel Wik

DOI: https://doi.org/10.1117/1.JATIS.9.4.048001

2023-10-27

Abstract:The large side aperture of the NuSTAR telescope for unfocused photons (so-called stray light) is a known source of rich astrophysical information. To support many studies based on the NuSTAR stray light data, we present a fully automatic method for determining detector area suitable for background analysis and free from any kind of focused X-ray flux. The method's main idea is `a trous' wavelet image decomposition, capable of detecting structures of any spatial scale and shape, which makes the method of general use. Applied to the NuSTAR data, the method provides a detector image region with the highest possible statistical quality, suitable for the NuSTAR stray light studies. We developed an open-source Python nuwavdet package, which implements the presented method. The package contains subroutines to generate detector image region for further stray light analysis and/or to produce a list of detector bad-flagged pixels for processing in the NuSTAR Data Analysis Software for conventional X-ray analysis.

Instrumentation and Methods for Astrophysics

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to solve how to automatically identify and remove any type of focused X - ray signal from the detector images of the NuSTAR telescope, thereby maximizing the clean detector area available for analysis. Specifically, this method focuses on dealing with so - called "stray light", that is, unfocused X - ray photons that directly irradiate the detector instead of photons focused through the optical system.

#### Main problems and challenges:
1. **Stray light background**: Due to the open geometry of the NuSTAR telescope, its focal plane module (FPM) can receive unfocused photons at an angle of 1 - 4 degrees away from the center of the field of view. These photons form a phenomenon called "stray light background", which provides rich astrophysical information.

2. **Interference of focused X - ray signals**: In order to effectively use the stray light background for scientific research, any focused X - ray signals that may interfere must be removed first. Traditional methods based on the point spread function (PSF) are inefficient and difficult to automate when dealing with a large amount of observational data.

3. **Automation and efficiency**: A fast, fully automatic and well - calibrated method is required to process large - scale data sets to ensure accurate separation of the stray light background and focused X - ray signals.

#### Proposed solutions:
The authors propose an image decomposition method based on wavelet transform (especially the 'à trous' algorithm). This method can detect structures of any spatial scale and shape and is very suitable for processing complex X - ray images. It is achieved through the following steps:

1. **Image pre - processing**: Generate the observed image in the DET1 coordinate system and perform image correction to deal with specific instrument artifacts, such as dead - time correction, detector gap filling, etc.

2. **Wavelet decomposition**: Use the 'à trous' digital wavelet transform algorithm to decompose the image into multiple scale layers. Each scale layer highlights structures with specific characteristic sizes.

3. **Conditional threshold filtering**: By setting an initial threshold \(t_{\text{max}}\) and a lower threshold \(t_{\text{min}}\), iteratively search for significant structures and expand the mask area to ensure that all focused X - ray signals are removed.

4. **Optimized mask combination**: By iteratively selecting wavelet mask combinations of different scales, find the best mask combination to maximize the clean detector area and optimize the statistical quality.

5. **Result verification**: This method was applied to process 6,573 NuSTAR observation data. The results show that this method can significantly increase the clean detector area available for stray light background research and provide high - quality statistical results.

Through this method, researchers can more effectively use the stray light background of the NuSTAR telescope for a wide range of astrophysical research, such as the cosmic X - ray background (CXB), diffuse hard X - ray emission, etc.