Elsevier

Acta Astronautica

Volume 183, June 2021, Pages 176-198
Acta Astronautica

Research paper
Detection and correlation of geosynchronous objects in NASA’s Wide-field Infrared Survey Explorer images

https://doi.org/10.1016/j.actaastro.2021.03.009Get rights and content

Highlights

  • Multi-band infrared measurements analysis of satellites and space debris.

  • Satellite detection algorithm requiring no prior knowledge of position of orbit parameters.

  • Infrared colour indices of geostationary satellites are similar to those of main-belt asteroids.

Abstract

In this paper, we present an algorithm developed to detect Earth-orbiting object streaks within astronomical images. The NASA Wide-field Infrared Survey Explorer (WISE) database of imagery was selected, in order to assess the viability of infrared wavelengths for the detection and analysis of satellite, rocket body and debris objects. The algorithm was designed to scan a high volume of images for the presence of streaks and then to correlate these to known Earth-orbiting objects without the use of a-priori knowledge of position or orbit parameters of the detected object. As part of the streak analysis process, the algorithm performed a photometric characterisation of the detected objects in all four infrared bands observed by the WISE satellite. The performance of the photometric analysis module was validated against known stellar magnitudes as well as the results of previous published research.

Introduction

There are currently no space-based infrared sensors conducting space surveillance operations that are aimed at observing active satellites and space debris objects in Earth orbit. That said, NASA launched the WISE spacecraft on 14 December 2009, into a Sun-synchronous Low Earth Orbit (LEO), with a mean altitude of 540 km and inclination of 97.5° [1]. The mission objective was to scan the entire sky in infrared wavelengths, to obtain insight into star formation, the existence of planets outside our solar system, and the origins of the universe. Over the course of its mission, the WISE space telescope serendipitously observed Earth orbiting satellites and space debris objects, commonly referred to as resident space objects (RSO). The expansive dataset of infrared images offered by the WISE mission therefore provides an ideal opportunity to detect and characterise RSOs, which is the premise of the work presented herein.

The WISE payload consists of a 40 cm telescope aperture, and three beam splitters that focus incident energy onto four separate focal plane arrays [1], [2]. These arrays take measurements in four infrared bands, W1-4, centred at 3.4, 4.6, 12 and 22μm [2] respectively. Exposures for all four bands start simultaneously but the exposure times differ: images in band W3 and W4 have 8.8 s exposure times, while W1 and W2 have a shorter 7.7 s exposure time due to a different readout process in the HgCdTe detector [3]. At the end of each image exposure, it takes 1.06 s to read each of the four arrays, and 1.1 s for the mirror to assume its new observational position. In practice, this means that an observational frame of four images (one for each band) is captured every 11 s.

Astronomical data collected from the WISE mission is managed by the Infrared Processing and Analysis Centre (IPAC), a division of NASA. Once downlinked to the ground, the raw images are converted into a FITS file along with supporting mission and image ancillary data that are included in the FITS header. Each FITS files is then calibrated for droop correction, linearisation, bias subtraction and flat fielding [1]. Source extraction as well as photometric and astrometric calibration are also performed, details of which are also included in the final FITS image header.

Finally, the data is available through the Infrared Science Archive (IRSA), which provides access to more than 20 billion astronomical measurements [5]. IRSA maintains data from all stages of the WISE mission, including operations prior to cryogenic depletion, W1 and W2 operations post-depletion, and the reactivated NEOWISE mission.

In 2015, Krezan et al. [6] used astronomical images collected by the NASA Wide-field Infrared Survey Explorer (WISE) space telescope to study the positional and size distributions of debris objects in, and near, the geostationary belt. Their work focused on the characterisation of the debris orbital location and their size distribution. Their analysis of serendipitous observations of these objects in the 4-band WISE images led them to propose a revised space debris population estimate as well as a new total collisional rate in the geostationary region. However, their work did not discuss what algorithm was used to detect the objects, if and how they were correlated to the U.S. space catalogue. Moreover, Krezan et al. did not provide photometric characteristics of the detected debris.

In a follow-up to the aforementioned work, Lee et al. [7], [8] analysed a subset of images collected by the WISE space telescope that were limited to a declination of ±15° thereby constraining the images to be studied to a region in proximity of the geostationary belt. The aim of their research was to conduct a multi-band infrared characterisation of geostationary satellites and rocket bodies in this region. In all, they detected and reported on the infrared photometric properties of 200 objects from 860 images out of the 82,208 that they retrieved from the NASA WISE image database. Contrary to work published by Krezen et al. [6], Lee et al. presented how they detected and correlated streaks that appeared in the WISE images.

The aim of the work presented in this paper was to develop a streak detection and correlation algorithm specifically customised to detect the serendipitous presence of artificial Earth objects in or near geosynchronous orbits in the NASA WISE imagery dataset. As part of this work, an evaluation of the algorithm performance was also conducted and compared to the work of Lee et al. [7], [8].

The work detailed in this paper was deemed necessary to enable future research on the analysis of using simultaneous observations of artificial Earth-orbiting objects in multiple infrared bands for the purpose of space object characterisation as well as to evaluate the utility of using the infrared portion of the electromagnetic spectrum for space surveillance purposes.

Section snippets

Algorithm overview

The algorithm developed and presented here is referred to as WISEstreakDET (WISE Streak Detection) throughout. The code primarily relies upon four concepts of image manipulation: background subtraction, thresholding, connectivity and image moments. These concepts, and those additional processes used to detect and analyse streaks, are summarised in the flowchart shown in Fig. 1. A pictorial representation of the algorithm is shown in Fig. 2, to provide context for the detailed explanation given

Image processing

In order to maximise the probability of detecting an RSO streak within a WISE image, a series of image manipulation processes were implemented. These processes seek to correct any defects present within images, and to distinguish between signal from sources and background noise.

Streak detection and photometry

In the previous step, pixels were assigned a value of one if above the set clipping threshold, or zero if below the threshold. This sought to discriminate pixels belonging to an object (point-source or streak) from those belonging to the background [16]. The next step is to identify individual objects within an image.

RSO correlation

Now that RSO streaks have been detected, and their photometric magnitudes determined, the final step is to match each streak to a known RSO — the process of correlation. A publicly available catalogue of RSOs is maintained by the US Space Command (USSPACECOM), viewable online at space-track.org [30]. The purpose of the correlation algorithm, described in detail throughout this section, is to match RSO data from the satellite catalogue to streaks detected by WISEstreakDET.

Streak detection and magnitude measurement issues

WISEstreakDET was configured to first detect streaks in band W3, in which detection had proved to be most successful. For each streak detected in W3, a masked image was created for all other bands that blanked out all pixels except for those of the detected streak location and surrounding area. This sought to ensure that magnitude for the same streak was measured in each band before progressing to the next streak.

An example of this is at Fig. 20, showing four streaks in band W4 of image frame

Algorithm validation and optical magnitude comparison

The accuracy and validity of WISEstreakDET was evaluated via two separate tests. The first test compared published star magnitudes to their magnitudes as measured with WISEstreakDET. The second test used WISEstreakDET to analyse a database of images examined in previous research, to compare the number of streaks detected, and their measured magnitudes.

Conclusion

Space domain awareness is becoming increasingly important for the safe space flight operations. The escalating number of resident space objects (RSO) – satellites, rocket bodies and debris – raises the probability of damage to space assets that are heavily relied upon in everyday life. RSOs are monitored and characterised by various sensors, predominantly exploiting the microwave, radio and optical segments of the electromagnetic spectrum. In this paper, we presented a streak detection and

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This publication makes use of data products from WISE, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, and NEOWISE, which is a project of the Jet Propulsion Laboratory/California Institute of Technology. WISE and NEOWISE are funded by the National Aeronautics and Space Administration.

References (45)

  • ProkopR. et al.

    A survey of moment-based techniques for unoccluded object representation and recognition

    Graph. Model and Image Process.

    (1992)
  • WrightE.L. et al.

    The wide-field infrared survey explorer (WISE): Mission description and initial on-orbit performance

    Astron. J.

    (2010)
  • LarsenM.F. et al.

    Wide-field infrared survey explorer science payload update

  • R.M. Cutri, E.L. Wright, T. Conrow, J. Bauer, D. Benford, H. Brandenburg, J. Dailey, P.R.M. Eisenhardt, T. Evans, S....
  • LévesqueM. et al.

    Image Processing Technique for Automatic Detection of Satellite StreaksTech. rep.

    (2007)
  • ALLWISE Source Catalogue, Caltech Infrared Processing and Analysis Center (IPAC). Retrieved from...
  • KrezanJ.M. et al.

    GEO Collisional risk assessment based on analysis of NASA-WISE data and modeling

  • LeeC.H. et al.

    Infrared photometry of GEO spacecraft with WISE

  • LeeC. et al.

    Distinguishing active box-wing and cylindrical geostationary satellites using IR photometry with NASA’s WISE spacecraft

  • GonzalezR. et al.

    Digital Image Processing using MATLAB

    (2004)
  • PesensonM. et al.

    High-dimensional data reduction, image inpainting and their astronomical applications

  • J. D‘Errico, inpaint_nans, MATLAB Central File Exchange,...
  • ChromeyF.R.

    To Measure the Sky: An Introduction to Observational Astronomy

    (2010)
  • LévesqueM. et al.

    Evaluation of the Iterative Method for Image Background Removal in Astronomical ImagesTech. rep.

    (2008)
  • WallaceB.

    Automated Streak Detection Using Image Segmentation and MomentsTech. rep.

    (2016)
  • ThorsteinsonS.

    Space Surveillance from a Microsatellite: Metric Observation Processing from NEOSSat

    (2017)
  • WuK. et al.

    Optimizing two-pass connected-component labeling algorithms

    Pattern Anal. Appl.

    (2009)
  • RosenfeldA.

    Connectivity in digital pictures

    J. ACM

    (1970)
  • HuM.

    Visual pattern recognition by moment invariants

    IRE Trans. Inf. Theory

    (1962)
  • PengM. et al.

    NIRFaceNet: A Convolutional neural network for near-infrared face identification

    Information

    (2016)
  • YauW. et al.

    Visual speech recognition using image moments and multiresolution wavelet images

  • PawlakM.

    Image Analysis by Moments : Reconstruction and Computational Aspects

    (2006)
  • Cited by (0)

    1

    Space Capability Officer, Royal Australian Air Force, M.Sc. (Physics, Space Operations) RMC Canada..

    2

    The views expressed in this article are personal views and should not be interpreted as an official position.

    3

    Adjunct Assistant Professor, Department of Applied Military Science..

    4

    Doctoral Candidate, Center for Imaging Science..

    5

    Research Professor (Emeritus), Department of Astronomy..

    View full text