High-Frequency Error Compensation for High-Resolution Satellite Stereo Images with Rough InSAR DSMs
Remote Sensing · 07.11.2026 · 18 (21) · 3836
DOI: 10.3390/rs18213836 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 07.11.2026. Verfasst von Tao Huang und 6 weiteren Autor:innen. Die Arbeit nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, the Second Batch of the Civil Aerospace Technology Pre-Research Project.
Abstract (Originalsprache)
High-precision digital surface models (DSMs) derived from optical stereo satellite data are essential for various applications. However, DSMs generated from high-resolution satellite imagery often contain high-frequency errors associated with calibration inaccuracies and attitude jitter. This study exploits differences in the error characteristics of optical stereo-derived and interferometric synthetic aperture radar (InSAR)-derived DSMs to detect, model, and compensate for these errors. Dense bundle adjustment first corrects low-frequency errors in the optical stereo image pair. Elevation differences between the resulting optical DSM and the InSAR DSM are then analyzed to identify coherent directional error patterns. An object-space compensation model based on median elevation-difference profiles is established, and the elevation corrections are transferred to image space to compensate for the stereo imagery. Experiments using a single Ziyuan-3 02 (ZY-3 02) stereo pair showed reduced stripe patterns associated with residual high-frequency errors. Using the 30 m Copernicus GLO-30 DSM as the elevation reference, the flight-direction median-difference root mean square error (RMSE) decreased from 0.43 to 0.22 m (48.8%). The scan-direction RMSE decreased from 0.51 to 0.25 m (51.0%). Both reductions were relative to the Quadratic adjustment. The image-space residual RMSE calculated from the fitted error curves decreased from 0.132 to 0.032 pixels after compensation. Independent evaluation against ALOS World 3D 30 m (AW3D30) also showed reduced high-frequency systematic distortions. The flight- and scan-direction median-difference RMSEs decreased from 0.58 to 0.35 m (39.7%) and from 0.78 to 0.37 m (52.6%), respectively.
Förderung
- National Natural Science Foundation of China
- the Second Batch of the Civil Aerospace Technology Pre-Research Project
Zitieren
APA
Huang, T., Liang, D., Cui, Z., Zhou, S., Zhu, X., Tang, H., & Pan, H. (2026). High-Frequency Error Compensation for High-Resolution Satellite Stereo Images with Rough InSAR DSMs. Remote Sensing, 18(21), 3836. https://doi.org/10.3390/rs18213836
Vancouver
Huang T, Liang D, Cui Z, Zhou S, Zhu X, Tang H, et al. High-Frequency Error Compensation for High-Resolution Satellite Stereo Images with Rough InSAR DSMs. Remote Sensing. 2026;18(21):3836. doi:10.3390/rs18213836
BibTeX
@article{Huang2026,
author = {Huang, Tao and Liang, Deyin and Cui, Zehua and Zhou, Shun and Zhu, Xiaoyong and Tang, Hong and Pan, Hongbo},
title = {High-Frequency Error Compensation for High-Resolution Satellite Stereo Images with Rough InSAR DSMs},
journal = {Remote Sensing},
year = {2026},
volume = {18},
number = {21},
pages = {3836},
doi = {10.3390/rs18213836}
}
EUVIMED