# Sieve Search Centroiding Algorithm for Star Sensors

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Sieve Search Algorithm

#### 2.1. Characteristics of PSF Gray Value Distribution

- the ROI matrix is nearly symmetric. For an $n\times n$ ROI matrix, let ${G}_{L}$ represent the n-element vector, consisting of gray values in the pixels along the leading diagonal ${d}_{L}$. The off-diagonals ${d}_{LU}$ and ${d}_{LD}$ are a pixels apart from ${d}_{L}$. The gray value vectors corresponding to ${d}_{LU}$ and ${d}_{LD}$ are ${G}_{LU}$ and ${G}_{LD}$, respectively. It can be observed that the value of the ${b}^{th}$ elements in ${G}_{LU}$ and ${G}_{LD}$ are almost similar, where $b\phantom{\rule{3.33333pt}{0ex}}\u03f5\phantom{\rule{3.33333pt}{0ex}}[1,n-a]$. In Figure 3, the off-diagonals ${d}_{LU}$ and ${d}_{LD}$ are two pixels apart from ${d}_{L}$. The grey value vectors ${G}_{L}=\left[2\phantom{\rule{3.33333pt}{0ex}}6\phantom{\rule{3.33333pt}{0ex}}29\phantom{\rule{3.33333pt}{0ex}}94\phantom{\rule{3.33333pt}{0ex}}149\phantom{\rule{3.33333pt}{0ex}}115\phantom{\rule{3.33333pt}{0ex}}43\phantom{\rule{3.33333pt}{0ex}}9\phantom{\rule{3.33333pt}{0ex}}3\right]$, ${G}_{LU}=\left[4\phantom{\rule{3.33333pt}{0ex}}19\phantom{\rule{3.33333pt}{0ex}}63\phantom{\rule{3.33333pt}{0ex}}104\phantom{\rule{3.33333pt}{0ex}}83\phantom{\rule{3.33333pt}{0ex}}33\phantom{\rule{3.33333pt}{0ex}}7\right]$, and ${G}_{LD}=\left[5\phantom{\rule{3.33333pt}{0ex}}22\phantom{\rule{3.33333pt}{0ex}}68\phantom{\rule{3.33333pt}{0ex}}104\phantom{\rule{3.33333pt}{0ex}}77\phantom{\rule{3.33333pt}{0ex}}28\phantom{\rule{3.33333pt}{0ex}}6\right]$. The values of ${b}^{th}$ elements in ${G}_{LU}$ and ${G}_{LD}$ are found to be comparable, where $b\phantom{\rule{3.33333pt}{0ex}}\u03f5\phantom{\rule{3.33333pt}{0ex}}[1,7]$. In other words, the pixels along the off-diagonals, equidistant from a diagonal element on both sides of the principal diagonals, have comparable gray value accumulation.
- as we transverse along the pixels at the same radial distance from the brightest pixel, the magnitude increases as we move closer to the principal diagonal pixels. Here, magnitude is referred to the gray value of the pixel. Let ${g}_{c}$ represent the gray value of a pixel ${p}_{c}$, along ${d}_{L}$, at some radial distance from the most illuminated pixel in the ROI. At the same radial distance, in the neighborhood of ${p}_{c}$, the gray value of the pixels along ${d}_{LU}$ and ${d}_{LD}$ are ${g}_{bu}$ and ${g}_{bd}$, respectively. Then, ${g}_{c}$ is marginally greater than ${g}_{bu}$ and ${g}_{bd}$. In Figure 3, the brightest pixel has an accumulated gray value of 149. Among the pixels at the same radial distance from it, the gray value at the pixel along the principal diagonal $\left\{43\right\}$ is marginally greater than that along the off-diagonal pixels $\{28,33\}$.

#### 2.2. Sieve Segmentation and Symmetry

#### 2.3. Characteristics of Sieve Magnitude

#### 2.4. Centroiding Based on Sieve Search Algorithm

- (i)
- An ROI of size not less than $5\times 5$ pixel window is created around the star image spot. The accumulated gray value distribution inside the ROI is used to populate the elements of the corresponding ROI matrix.
- (ii)
- The ROI matrix is segregated into square sub-matrices called sieves. The degree of symmetry and magnitude of each sieve ${S}_{i}$ in the ROI is evaluated and stored as ${\varphi}_{i}$ and ${\mu}_{i}$, respectively, where $i\phantom{\rule{3.33333pt}{0ex}}\u03f5\phantom{\rule{3.33333pt}{0ex}}1,\dots ,{k}^{2}$. The sieves are rotated by ${90}^{\circ}$ clockwise. The corresponding ${\varphi}_{i[+{90}^{\circ}]}$ and ${\mu}_{i[+{90}^{\circ}]}$ are determined.
- (iii)
- The sieve probability index $P\left({S}_{i}\right)$, evaluated for each sieve ${S}_{i}$ in the ROI according to Equation (7), gives the probability of a sieve to containing the centroid location of the image spot. This is depicted in Figure 6.$${\rho}_{i}=({\varphi}_{i}+{\varphi}_{i[+{90}^{\circ}]})\xb7({\mu}_{i}+{\mu}_{i[+{90}^{\circ}]}),\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}P\left({S}_{i}\right)=\frac{{\rho}_{i}}{max\left({\rho}_{i}\right)}$$
- (iv)
- For every sieve ${S}_{i}$ in the ROI, the value of its degree of symmetry–magnitude product ${\rho}_{i}$ is associated with each pixel contained in it. Consequently, for a pixel j, the value of ${\rho}_{i}$, resulting from the various sieves it constitutes, adds up, given by,$${\psi}_{j}=\sum _{i\u03f5U}{\rho}_{i}$$$$P\left(j\right)=\frac{{\psi}_{j}}{max\left({\psi}_{j}\right)}$$
- (v)
- Let ${h}_{j}$ and ${v}_{j}$ correspond to the horizontal and vertical coordinates of pixel j. The value of the centroid is computed as,$${h}_{c}=\frac{{\sum}_{j\u03f5W}{h}_{j}\xb7{\psi}_{j}}{{\sum}_{j\u03f5W}{\psi}_{j}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}\phantom{\rule{3.33333pt}{0ex}}{v}_{c}=\frac{{\sum}_{j\u03f5W}{v}_{j}\xb7{\psi}_{j}}{{\sum}_{j\u03f5W}{\psi}_{j}}$$

#### 2.5. Optimal Sieve Size

## 3. Materials and Methods

#### 3.1. Simulation of Star Images

#### 3.2. Simulation of Noise Process

#### 3.3. Selection of Attributes

- the radius of the Gaussian spread in both directions, ${\sigma}_{PSF}=0.75$.
- the size of the star image pixel window was $3\times 3$.
- the average level of noise corresponding to a star image spot was restricted to 7% of the accumulated gray value in the most illuminated pixel of that star image.

#### 3.4. Particulars of Experiments

## 4. Results and Analysis

#### 4.1. Sensitivity of Centroiding Accuracy to Various Parameters

#### 4.1.1. Brightness

#### 4.1.2. Location of Star Spot Inside a Pixel

#### 4.1.3. Noise Floor

#### 4.1.4. Gaussian Spread

#### 4.2. Special Cases: Performance Evaluation

#### 4.2.1. Non-Uniform PSF

#### 4.2.2. Stuck Pixel Noise

#### 4.2.3. Optical-Double Stars

#### 4.3. Execution Time

## 5. Conclusions

## Author Contributions

## Funding

## Informed Consent Statement

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Fallon, L., III. Star sensors. In Spacecraft Attitude Determination and Control; Wertz, J.R., Ed.; D. Reidel Publishing Company: Dordrecht, The Netherlands, 1978; p. 186. ISBN 978-94-009-9907-7. [Google Scholar]
- Liebe, C.C. Accuracy performance of star trackers—A tutorial. IEEE Trans. Aerosp. Electron. Syst.
**2002**, 38, 587–599. [Google Scholar] [CrossRef] - Samaan, M.A. Toward Faster and More Accurate Star Sensors Using Recursive Centroiding and Star Identification. Ph.D. Thesis, Texas A&M University, Austin, TX, USA, 2003. [Google Scholar]
- Wang, H.; Xu, E.; Li, Z.; Jingjin, L.; Quin, T. Gaussian analytic centroiding method of star image of star tracker. Adv. Space Res.
**2015**, 56, 2196–2205. [Google Scholar] [CrossRef] - Huffman, K.M.; Sedwick, R.J.; Stafford, J.; Pavaerill, J.; Seng, W. Designing star trackers to meet micro-satellite requirements. In Proceedings of the International Conference on Space Operations (SpaceOps2006), Rome, Italy, 19–23 June 2006; pp. 1–29. [Google Scholar] [CrossRef] [Green Version]
- Baker, K.L.; Moallem, M.M. Iteratively weighted centroiding for Shack- Hartmann wave-front sensors. Opt. Express
**2007**, 15, 1–13. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Shortis, M.; Clarke, T.; Short, T. A Comparison of some techniques for the subpixel location of discrete target images. SPIE Videometrics III
**1994**, 2350, 239–250. [Google Scholar] [CrossRef] - Sheppard, C.J.R. Microscopy overview. In Encyclopedia of Modern Optics; Guenther, R.D., Steel, D.G., Bayvel, L., Eds.; Elsevier: Oxford, UK, 2004; pp. 61–68. ISBN 978-0-12-369395-2. [Google Scholar]
- Schowengerdt, R.A. Sensor models. In Remote Sensing; Schowengerdt, R.A., Ed.; Academic Press: Cambridge, MA, USA, 2007; Chapter 3; pp. 75–126. ISBN 978-0-12-369407-2. [Google Scholar]
- Airy, G.B. On the diffraction of an object-glass with circular aperture. Trans. Cambridge Philos. Soc.
**1835**, 5, 283–291. [Google Scholar] - Katake, A.B. Modeling, Image Processing and Attitude Estimation of High Speed Star Sensors. Ph.D. Thesis, Texas A&M University, Austin, TX, USA, 2006. [Google Scholar]
- Zhang, B.; Zerubia, J.; Olivo-Marin, J. Gaussian approximations of fluorescence microscope point-spread function models. Appl. Opt.
**2007**, 46, 1819–1829. [Google Scholar] [CrossRef] [PubMed] - Mortari, D.; Bruccoleri, C.; La Rosa, S.; Junkins, J.L. CCD Data Processing Improvements for star Cameras. In Proceedings of the International Conference on Dynamics and Control of System and Structures in Space, Denver, CO, USA, 22–25 April 2002. [Google Scholar]
- Tremsin, A.S.; Vallerga, J.V.; Siegmund, O.H.W.; Hull, J.S. Centroiding algorithms and spatial resolution of photon counting detectors with cross strip anodes. Proc. SPIE
**2003**, 5164, 1–12. [Google Scholar] [CrossRef] - Yang, J.; Liang, B.; Zhang, T.; Song, J. A novel systematic error compensation algorithm based on least squares support vector regression for star sensor image centroid estimation. Sensors
**2011**, 11, 7341–7363. [Google Scholar] [CrossRef] [Green Version] - Wei, X.; Xu, J.; Li, J. S-curve centroiding error correction for star sensor. Acta Astronaut.
**2014**, 99, 231–241. [Google Scholar] [CrossRef] - Delabie, T.; Schutter, J.D.; Vandenbussche, B. An Accurate and efficient Gaussian fit centroiding algorithm for star trackers. J. Astronaut.
**2014**, 61, 60–84. [Google Scholar] [CrossRef] - Wan, X.; Wang, G.; Wei, X.; Li, J.; Zhang, G. Star centroiding based on fast Gaussian fitting for star sensors. Sensors
**2018**, 18, 2836. [Google Scholar] [CrossRef] [Green Version] - Fialho, M.A.A.; Mortari, D. Theoretical limits of star sensor accuracy. Sensors
**2019**, 19, 5355. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Asadnezhad, M.; Eslamimajd, A.; Hajghassem, H. Optical system design of star sensor and stray light analysis. J. Eur. Opt. Soc.
**2018**, 14, 1–11. [Google Scholar] [CrossRef] [Green Version] - Wei, X.; Wen, D.; Song, Z.; Xi, J.; Zhang, W.; Liu, G.; Li, Z. A systematic error compensation method based on an optimized extreme learning machine for star sensor image centroid estimation. Appl. Sci.
**2019**, 9, 4751. [Google Scholar] [CrossRef] [Green Version] - Pong, C.M.; Smith, M.W. Camera modeling, centroiding performance, and geometric camera calibration on ASTERIA. In Proceedings of the IEEE Aerospace Conference 2019, Big Sky, MT, USA, 2–9 March 2019; pp. 1–17. [Google Scholar] [CrossRef]
- Wei, X.; Tan, W.; Li, J.; Zhang, G. Exposure time optimization for highly dynamic star trackers. Sensors
**2014**, 14, 4914–4931. [Google Scholar] [CrossRef] [Green Version] - Di1, P.; Qi1, Z.; Weizhi, Q.; Bing, L.; Pingchuan, R. Star spot centroid extraction method in high dynamic condition based on difference hash algorithm. In Proceedings of the 12th Asia Conference on Mechanical and Aerospace Engineering (ACMAE 2021), Nanjing, China, 29–31 December 2021; pp. 1–6. [Google Scholar] [CrossRef]
- Cao, Z.; Wang, G.; Wei, X. Improving star centroiding accuracy in stray light base on background estimation. In Proceedings of the IEEE 9th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON 2018), Vancouver, ON, Canada, 1–3 November 2018; pp. 513–518. [Google Scholar] [CrossRef]
- He, Y.; Wang, H.; Feng, L.; You, S.; Lu, J.; Jiang, W. Centroid extraction algorithm based on grey-gradient for autonomous star sensor. Optik
**2019**, 194, 162932. [Google Scholar] [CrossRef] - Zhang, Y.; Jiang, J.; Zhang, G.; Lu, Y. Accurate and robust synchronous extraction algorithm for star centroid and nearby celestial body edge. IEEE Access
**2019**, 7, 126742–126752. [Google Scholar] [CrossRef] - Samed, A.L.; Karagoz, I.; Dogan, A. An Improved star detection algorithm using a combination of statistical and morphological image processing techniques. In Proceedings of the IEEE Applied Imagery Pattern Recognition Workshop (AIPR 2018), Washington, DC, USA, 9–11 October 2022; pp. 1–6. [Google Scholar] [CrossRef]
- Zhu, H.; Xie, J.; Tang, X.; Jia, D.; Sun, G. Stellar map centroid positioning based on dark channel denoising and feasibility of jitter detection on ZiYuan3 satellite platform. J. Appl. Rem. Sens.
**2021**, 15, 016519. [Google Scholar] [CrossRef] - Lu, K.; Liu, E.; Zhao, R.; Zhang, H.; Tian, H. Star sensor denoising algorithm based on edge protection. Sensors
**2021**, 21, 5255. [Google Scholar] [CrossRef] - Mu, Z.; Wang, J.; He, X.; Wei, Z.; He, J.; Zhang, L.; Lv, Y.; He, D. Restoration method of a blurred star image for a star sensor under dynamic conditions. Sensors
**2019**, 19, 4127. [Google Scholar] [CrossRef] [Green Version] - Erlank, A.O. Development of cubeStar: A cubeSat-Compatible Star Tracker. Master’s Thesis, Stellenbosch University, Stellenbosch, South Africa, 2013. [Google Scholar]
- Ofodile, I.; Kutt, J.; Kivastik, J.; Nigol, M.K.; Parelo, A.; Ilbis, E.; Ehrpais, H.; Slavinskis, A. ESTCube-2 attitude determination and control: Step towards interplanetary cubeSats. In Proceedings of the IEEE Aerospace Conference 2019, Big Sky, MT, USA, 2–9 March 2019; pp. 1–12. [Google Scholar] [CrossRef]
- Kopacz, J.R.; Herschitz, R.; Roney, J. Small satellites an overview and assessment. Acta Astronaut.
**2020**, 170, 93–105. [Google Scholar] [CrossRef] - Qian, X.; Yu, H.; Chen, S. A global-shutter centroiding measurement CMOS image sensor with star region SNR improvement for star trackers. IEEE Trans. Circuits Syst. Video Technol.
**2016**, 26, 1555–1562. [Google Scholar] [CrossRef] - Bao, J.; Zhan, H.; Sun, T.; Fu, S.; Xing, F.; You, Z. A window-adaptive centroiding method based on energy iteration for spot target localization. IEEE Trans. Instrum. Meas.
**2022**, 71, 1–13. [Google Scholar] [CrossRef] - Piotrowski, L.; Batsch, T.; Czyrkowski, H.; Cwiok, M.; Dabrowski, R.; Kasprowicz, G.; Majcher, A.; Majczyna, A.; Małek, K.; Mankiewicz, L.; et al. PSF modelling for very wide-field CCD astronomy. Astron. Astrophys.
**2013**, 551, 1–15. [Google Scholar] [CrossRef] [Green Version] - Bodewig, E. Measures of the magnitude of a matrix. In Matrix Calculus; Bodewig, E., Ed.; Elsevier: Amsterdam, The Netherlands, 1959; p. 43. ISBN 978-1-4832-3214-0. [Google Scholar]
- STAR 1000 Detailed Specification; APS-FF-DU-03-004; Fillfactory Image Sensors: Belgium. 2006. Available online: https://escies.org/download/webDocumentFile?id=62275 (accessed on 22 February 2023).
- Rabie, T. Adaptive hybrid mean and median filtering of high-ISO long-exposure sensor noise for digital photography. J. Electron. Imaging
**2004**, 13, 264–277. [Google Scholar] [CrossRef] - Buchheim, R.K. CCD Double Star Observations at Altimira Observatory: Spring 2008. J. Double Star Obs.
**2008**, 3, 103–110. [Google Scholar] - Buchheim, R.K. CCD Measurements of Visual Double Stars. In Proceedings of the 27th Annual Symposium on Telescope Science, Big Bear Lake, CA, USA, 20–22 May 2008; pp. 13–23. [Google Scholar]

**Figure 2.**Pixel intensity of star image spot on a sensor array. The actual centroid of this star image is designated at $(804.35,182.87)$. The pixel closest to the centroid is the brightest and generates 1491 photons. (

**a**) The point spread function is (

**b**) The gray-scale accumulation inside the ROI.

**Figure 4.**The degree of symmetry for all sieves ${S}_{i}$ and ${S}_{i[+{90}^{\circ}]}$ in the ROI. (

**a**) $\varphi $ of all sieves S in the ROI. $\varphi $ peaks for sieves at index numbers $u=[1,\phantom{\rule{3.33333pt}{0ex}}9,\phantom{\rule{3.33333pt}{0ex}}17,\phantom{\rule{3.33333pt}{0ex}}25,\phantom{\rule{3.33333pt}{0ex}}33,\phantom{\rule{3.33333pt}{0ex}}41,\phantom{\rule{3.33333pt}{0ex}}49]$; (

**b**) ${\varphi}_{[+{90}^{\circ}]}$ of all sieves ${S}_{[+{90}^{\circ}]}$ in the ROI. ${\varphi}_{[+{90}^{\circ}]}$ peaks at $v=[7,\phantom{\rule{3.33333pt}{0ex}}13,\phantom{\rule{3.33333pt}{0ex}}19,\phantom{\rule{3.33333pt}{0ex}}25,\phantom{\rule{3.33333pt}{0ex}}31,\phantom{\rule{3.33333pt}{0ex}}37,\phantom{\rule{3.33333pt}{0ex}}43]$.

**Figure 9.**The optimal size of the sieve. (

**a**) Transient and steady-state centroiding errors for sieve size $m=2,3$ and 6; (

**b**) Comparison of various sieve sizes, based on the centroiding error and execution time.

**Figure 11.**The variation of centroiding error with respect to the changes in the brightness level of the star spot. The SSA algorithm was comparatively stable throughout the operation regime.

**Figure 12.**The variation of centroiding errors to changes in the location of the star image spot inside a pixel for various centroiding algorithms. The variation centroiding accuracy, due to S-curve errors, was minimal for SSA, compared to COM and IWCOG. (

**a**) GAC; (

**b**) COM; (

**c**) IWCOG; (

**d**) GBF, G3P and FGF; (

**e**) SSA.

**Figure 13.**The variation of centroiding error with the changes in noise level. The noise robustness of SSA was found to be equivalent to that of FGF and G3P.

**Figure 14.**The variation of centroiding error with the changes in the Gaussian radius. The centroiding error in SSA was high until ${\sigma}_{PSF}\approx 0.9$. Beyond this, as the Gaussian radius increased, the centroiding accuracy of SSA coincided with those of GBF, G3P, and FGF.

**Figure 15.**The special cases encountered by the star sensor camera during its operational lifespan. (

**a**) Non-uniform PSF; (

**b**) Stuck-pixel noise; (

**c**) Optical double stars.

**Figure 16.**The variation of centroiding error with an increase in ${\sigma}_{PSF-Y}$ from 0.5 to 1.5. For ${\sigma}_{PSF-Y}>0.9$, the centroiding accuracy of SSA was comparable with those of GBF, G3P, and FGF.

**Figure 17.**Sensitivity of centroiding accuracy to the location of stuck-pixel noise in the ROI. The magnitude of error in SSA was insignificant compared to the other algorithms, except for GBF.

**Figure 18.**The centroiding error in resolving the visual magnitude barycenter of an optical double star pair. The SSA follows the actual centroid closely, resulting in minimum centroiding error.

**Figure 19.**Comparison of relative real-time computer-simulated execution time for various centroiding algorithms. The gray-scale algorithms COM, IWCOG, and GAC were found to have the lowest execution time. SSA was found to require considerably less execution time in comparison with the curve fitting algorithms.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Karaparambil, V.C.; Manjarekar, N.S.; Singru, P.M.
Sieve Search Centroiding Algorithm for Star Sensors. *Sensors* **2023**, *23*, 3222.
https://doi.org/10.3390/s23063222

**AMA Style**

Karaparambil VC, Manjarekar NS, Singru PM.
Sieve Search Centroiding Algorithm for Star Sensors. *Sensors*. 2023; 23(6):3222.
https://doi.org/10.3390/s23063222

**Chicago/Turabian Style**

Karaparambil, Vivek Chandran, Narayan Suresh Manjarekar, and Pravin Madanrao Singru.
2023. "Sieve Search Centroiding Algorithm for Star Sensors" *Sensors* 23, no. 6: 3222.
https://doi.org/10.3390/s23063222