From Fixed to Automated Mirror Arrays: Advancing Multi-Satellite Image Registration Toward Operational Harmonization
MUHAMMAD DANIEL IMAN BIN HUSSAIN (a*), MASAHIKO NAGAI (a), VAIBHAV KATIYAR (b)

a) Graduate School of Sciences and Technology for Innovation, Yamaguchi University, Japan
*danieliman9897[at]gmail.com
b) New Space Intelligence Inc., Japan

Abstract

Accurate alignment of satellite imagery across different sensors and spectral bands is critical for reliable multi-source remote sensing applications. Building on our previous work that identified a Fourier-based hybrid co-registration method-combining mirror arrays with local refinement-as the most effective strategy, this study focuses on extending the method to support high-accuracy registration across multiple satellites and imaging conditions. We evaluate a newly developed automated mirror array system that uses TLE-based satellite tracking and dynamic signal adjustment via variable mirror sizes to accommodate spatial resolutions from 15 m to 0.5 m. Compared to the previously used fixed mirror arrays, the automated system requires no manual setup or repositioning and ensures more consistent, real-time alignment with satellite overpasses. The hybrid co-registration method is applied to imagery from Sentinel-2, PlanetScope, and GRUS satellites, targeting both inter-satellite co-registration and intra-satellite band-to-band registration. We quantitatively assess improvements in registration accuracy using RMSE and CE90 metrics, and qualitatively validate results using the Global Reference Image (GRI). A practical case study is demonstrated to illustrate the impact of improved registration on downstream analysis, where misalignment between high and medium resolution images may lead to inaccurate change detection. Results show that the automated mirror array improves geometric consistency across sensors and enhances the interpretability of stacked multi-temporal data. The findings underscore the potential operational advantages of automated mirror arrays as scalable ground references for the harmonization of multi-satellite imagery, facilitating more accurate Earth observation and downstream applications such as environmental monitoring and urban analysis. This work also lays the foundation for future research into integrated radiometric calibration and automated calibration pipelines for emerging microsatellite constellations.

Keywords: Satellite image registration, Mirror array, Multi-sensor harmonization, microsatellites

Topic: Topic D: Geospatial Data Integration