儘管推掃式光學衛星感測器具備高訊噪比之優勢，惟其逐行掃描機制使影像品質對平台微振敏感。在衛星運行期間，內部元件運作所引發的微振常導致感測器視軸發生高頻抖動，進而造成影像產生幾何畸變與模糊。鑑於現有監督式深度學習方法面臨帶有標記訓練資料匱乏，且既有簡化模擬難以反映真實成像幾何之雙重挑戰，本研究提出一套整合「嚴密成像模擬」與「端對端深度學習」之衛星影像微振補償框架。

在資料生成策略方面，本研究參照福爾摩沙衛星八號 (Formosat-8) 之成像參數，結合台灣地區 Google Map 正射影像與 AW3D30 數值高程模型 (DEM)，建構了一套具備嚴密幾何物理機制之推掃式影像模擬系統。該系統可藉由調控振動頻率與振幅參數，引入多樣化的姿態擾動，並在精確考量軌道動力學與地形起伏的情境下模擬成像過程，最終生成具備幾何一致性的退化影像與其真值資料，有效克服深度學習訓練樣本稀缺之瓶頸。

在模型架構設計上，本研究提出了一種端對端深度學習模型 EJDIR-Net。鑑於推掃式成像獨有的時間序列特性，本架構採用非等向性池化 (Anisotropic Pooling) 策略，旨在特徵提取階段最大程度保留沿軌方向的高頻振動資訊。此外，本研究具創新性地在類神經網路瓶頸層導入了姿態參數監督機制，透過姿態參數引導模型精確捕捉並解耦微振訊號。

實驗結果證實EJDIR-Net 的表現顯著優於現有模型。經補償後，影像平均峰值訊噪比 (PSNR) 由 24.95 dB 提升至 34.48 dB，結構相似性指標 (SSIM) 由 0.785 提升至 0.966 ；姿態估測平均絕對誤差 (MAE) 僅為 0.0014 角秒 。綜合而言，本研究不僅能精確偵測高頻抖動，亦能顯著改善影像清晰度與幾何一致性，為高解析度衛星影像品質優化提供具實用價值之解決方案。

The advancement of remote sensing technology has increasingly relied on high-resolution optical satellites equipped with linear pushbroom sensors. While these sensors offer superior signal-to-noise ratios and spatial resolution, their continuous line-by-line scanning mechanism introduces a critical vulnerability to satellite platform micro-vibrations. These disturbances manifest as high-frequency jitter in the optical line-of-sight (LOS), resulting in geometric distortions and image blurring. To address the limitations of existing deep learning methods, particularly the scarcity of paired training data and the oversimplification of training data generation, this research proposes a comprehensive framework integrating a pushbroom sensor simulator and an End-to-End Jitter Detection and Image Restoration Network (EJDIR-Net). The proposed simulator constructs a physically accurate dataset by modeling the rigorous orbital imaging geometry and terrain interactions. The EJDIR-Net architecture features a novel anisotropic pooling strategy to preserve high-frequency jitter information and introduces a parameter supervision mechanism to facilitate model convergence. Experimental results validate that EJDIR-Net significantly outperforms state-of-the-art methods, improving the Peak Signal-to-Noise Ratio (PSNR) to 34.48 dB and achieving a jitter estimation Mean Absolute Error (MAE) of 0.0014 arcseconds, demonstrating robust generalization on real-world imagery.

Antun, V., Renna, F., Poon, C., Adcock, B., and Hansen, A. C. (2020). On instabilities of deep learning in image reconstruction and the potential costs of AI. Proceedings of the National Academy of Sciences, 117(48):30088–30095.
Biquard, M., Chabert, M., Genin, F., Latry, C., and Oberlin, T. (2025). PG-DPIR: An efficient plug-and-play method for high-count Poisson-Gaussian inverse problems. arXiv preprint arXiv:2504.10375.
Cao, H., Tao, P., Li, H., and Shi, J. (2019). Bundle adjustment of satellite images based on an equivalent geometric sensor model with digital elevation model. ISPRS Journal of Photogrammetry and Remote Sensing, 156:169–183.
Chen, L.-C., Papandreou, G., Schroff, F., and Adam, H. (2017). Rethinking atrous convolution for semantic image segmentation. arXiv preprint arXiv:1706.05587.
Cobb, R. G., Sullivan, J. M., Das, A., Davis, L. P., Hyde, T. T., Davis, T., Rahman, Z. H., and Spanos, J. T. (1999). Vibration isolation and suppression system for precision payloads in space. Smart Materials and Structures, 8(6):798.
Davis, L. P., Wilson, J., Jewell, R., and Roden, J. (1986). Hubble space telescope reaction wheel assembly vibration isolation system. NASA Marshall Space Flight Center, Huntsville, Alabama, 9.
Delevit, J., Greslou, D., Amberg, V., Dechoz, C., de Lussy, F., Lebegue, L., Latry, C., Artigues, S., and Bernard, L. (2012). Attitude assessment using Pleiades-HR capabilities. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 39:525–530.
Dennehy, C. J. and Alvarez-Salazar, O. S. (2019). A survey of the spacecraft line-of-sight jitter problem. In AAS Annual Guidance and Control Conference.
Diakogiannis, F. I., Waldner, F., Caccetta, P., and Wu, C. (2020). ResUNet-a: A deep learning framework for semantic segmentation of remotely sensed data. ISPRS Journal of Photogrammetry and Remote Sensing, 162:94–114.
Dong, C., Loy, C. C., He, K., and Tang, X. (2014). Learning a deep convolutional network for image super-resolution. In European conference on computer vision, pages 184–199. Springer.
Fan, C., Wang, M., Zhao, W., Yang, B., Jin, S., and Pan, J. (2016). A compensation modeling method for time-varying systematic error of high-resolution optical satellite image. Acta Optica Sinica, 36(12):308–315.
Foi, A., Trimeche, M., Katkovnik, V., and Egiazarian, K. (2008). Practical Poissonian-Gaussian noise modeling and fitting for single-image raw-data. IEEE Transactions on Image Processing, 17(10):1737–1754.
Gao, X., Wang, G., Guan, S., et al. (2021). Micro-vibration attenuation design and verification for GFDM-1 satellite. Spacecraft Environment Engineering, 30(3):76–85.
Guan, X., Wang, G.-y., Cao, D.-j., Tang, S.-f., Chen, X., Liang, L., and Zheng, G.-t. (2017). Experimental demonstration of 1.5 Hz passive isolation system for precision optical payloads. In International Conference on Space Optics—ICSO 2010, volume 10565, pages 393–399. SPIE.
Hindle, T., Davis, T., and Fischer, J. (2007). Isolation, pointing, and suppression (IPS) system for high-performance spacecraft. In Industrial and Commercial Applications of Smart Structures Technologies 2007, volume 6527, pages 37–48. SPIE.
Iwasaki, A. (2011). Detection and Estimation Satellite Attitude Jitter Using Remote Sensing Imagery. In Hall, J., editor, Advances in Spacecraft Technologies, chapter 13. IntechOpen, London.
Jiang, Y.-h., Zhang, G., Tang, X., Li, D., and Huang, W.-c. (2014). Detection and correction of relative attitude errors for ZY1-02C. IEEE Transactions on Geoscience and Remote Sensing, 52(12):7674–7683.
Kawak, B. (2017). Development of a low-cost, low micro-vibration CMG for small agile satellite applications. Acta Astronautica, 131:113–122.
Kim, M.-J., Kim, M.-S., and Shin, S. Y. (1995). A general construction scheme for unit quaternion curves with simple high order derivatives. In Proceedings of the 22nd annual conference on Computer graphics and interactive techniques, pages 369–376.
Latry, C., Fourest, S., and Thiebaut, C. (2012). Restoration technique for Pleiades-HR panchromatic images. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 39:555–560.
LeCun, Y., Bengio, Y., and Hinton, G. (2015). Deep learning. Nature, 521(7553):436–444.
Ledig, C., Theis, L., Huszár, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z., et al. (2017). Photo-realistic single image super-resolution using a generative adversarial network. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4681–4690.
Li, M., Zhang, Y., Wang, Y., Hu, Q., and Qi, R. (2019). The pointing and vibration isolation integrated control method for optical payload. Journal of Sound and Vibration, 438:441– 456.
Maglione, P. (2016). Very high resolution optical satellites: An overview of the most commonly used. American Journal of Applied Sciences, 13(1):91.
Mosier, G. E., Howard, J. M., Johnston, J. D., Parrish, K. A., Hyde, T. T., McGinnis, M. A., Bluth, A. M., Kim, K., and Ha, K. Q. (2004). The role of integrated modeling in the design and verification of the James Webb Space Telescope. In Space Systems Engineering and Optical Alignment Mechanisms, volume 5528, pages 96–107. SPIE.
Mumtaz, R. and Palmer, P. (2013). Attitude determination by exploiting geometric distortions in stereo images of DMC camera. IEEE Transactions on Aerospace and Electronic Systems, 49(3):1601–1625.
Nah, S., Kim, T. H., and Lee, K. M. (2017). Deep multi-scale convolutional neural network for dynamic scene deblurring. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3883–3891.
Nguyen, M. H., de Vieilleville, F., and Weiss, P. (2024). DeepVibes: Correcting Micro-Vibrations in Satellite Imaging with Pushbroom Cameras. IEEE Transactions on Geoscience and Remote Sensing, 62:1–9.
Pan, H., Tao, C., and Zou, Z. (2016). Precise georeferencing using the rigorous sensor model and rational function model for ZiYuan-3 strip scenes with minimum control. ISPRS Journal of Photogrammetry and Remote Sensing, 119:259–266.
Pan, J., Che, C., Zhu, Y., and Wang, M. (2017). Satellite Jitter Estimation and Validation Using Parallax Images. Sensors, 17(1).
Pendergast, K. J. and Schauwecker, C. J. (1998). Use of a passive reaction wheel jitter isolation system to meet the advanced X-ray astrophysics facility imaging performance requirements. In Space telescopes and instruments V, volume 3356, pages 1078–1094. SPIE.
Poli, D. and Toutin, T. (2012). Review of developments in geometric modelling for high resolution satellite pushbroom sensors. The Photogrammetric Record, 27(137):58–73.
Shi, Y., Tong, X., Wen, J., Zhao, H., Ying, X., and Zha, H. (2021). Position-aware and symmetry enhanced GAN for radial distortion correction. In 2020 25th International Conference on Pattern Recognition (ICPR), pages 1701–1708. IEEE.
Shin, D., Pollard, J., and Muller, J.-P. (1997). Accurate geometric correction of ATSR images. IEEE Transactions on Geoscience and Remote Sensing, 35(4):997–1006.
Song, X., Fang, Z., and Pan, J. (2017). Image deblurring method for satellite platform caused by jitter. Journal on Communications, 38(Z1):186–192.
Takaku, J. and Tadono, T. (2010). High resolution dsm generation from alos prism-processing status and influence of attitude fluctuation. In 2010 IEEE International Geoscience and Remote Sensing Symposium, pages 4228–4231. IEEE.
Tang, X., Xie, J., Zhu, H., and Mo, F. (2020). Overview of Earth Observation Satellite Platform Microvibration Detection Methods. Sensors, 20(3).
Tao, X., Gao, H., Shen, X., Wang, J., and Jia, J. (2018). Scale-recurrent network for deep image deblurring. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 8174–8182.
Tian, C., Xu, Y., and Zuo, W. (2020). Image denoising using deep CNN with batch renormalization. Neural Networks, 121:461–473.
Tong, X., Ye, Z., Xu, Y., Tang, X., Liu, S., Li, L., Xie, H., Wang, F., Li, T., and Hong, Z. (2014). Framework of jitter detection and compensation for high resolution satellites. Remote Sensing, 6(5):3944–3964.
Toutin, T. (2004). Review Article: Geometric Processing of Remote Sensing Images: Models, Algorithms and Methods. International Journal of Remote Sensing, 25(10):1893– 1924.
Urasaki, C., Zhu, F., Bottom, M., Nunes, M., and Walk, A. (2024). Jitter Characterization of the HyTI Satellite. In 2024 IEEE Aerospace Conference, pages 1–16. IEEE.
Uuemaa, E., Ahi, S., Montibeller, B., Muru, M., and Kmoch, A. (2020). Vertical accuracy of freely available global digital elevation models (ASTER, AW3D30, MERIT, TanDEM-X, SRTM, and NASADEM). Remote Sensing, 12(21):3482.
Wang, M., Fan, C., Pan, J., Jin, S., and Chang, X. (2017). Image jitter detection and compensation using a high-frequency angular displacement method for Yaogan-26 remote sensing satellite. ISPRS Journal of Photogrammetry and Remote Sensing, 130:32–43.
Wang, P., An, W., Deng, X., Yang, J., and Sheng, W. (2015). A jitter compensation method for spaceborne line-array imagery using compressive sampling. Remote Sensing Letters, 6(7):558–567.
Wang, Z., Zhang, Z., Dong, L., and Xu, G. (2021). Jitter detection and image restoration based on generative adversarial networks in satellite images. Sensors, 21(14):4693.
Wittig, M. E., Van Holtz, L., Tunbridge, D. E. L., and Vermeulen, H. C. (1990). In-orbit measurements of microaccelerations of ESA’s communication satellite OLYMPUS. In Free-Space Laser Communication Technologies II, volume 1218, pages 205–214. SPIE.
Xiao, J., Lyu, Z., Zhang, C., Ju, Y., Shui, C., and Lam, K.-M. (2024). Towards progressive multi-frequency representation for image warping. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2995–3004.
Xing, W., Tuo, W., Li, X., Wang, T., and Yang, C. (2024). Micro-vibration suppression and compensation techniques for in-orbit satellite: A review. Chinese Journal of Aeronautics, 37(9):1–19.
Ye, Z., Xu, Y., Zheng, S., Tong, X., Xu, X., Liu, S., Xie, H., Liu, S., Wei, C., and Stilla, U. (2020). Resolving time-varying attitude jitter of an optical remote sensing satellite based on a time-frequency analysis. Optics Express, 28(11):15805–15823.
Zhang, G. and Guan, Z. (2018). High-frequency attitude jitter correction for the Gaofen-9 satellite. The Photogrammetric Record, 33(162):264–282.
Zhang, K., Ren, W., Luo, W., Lai, W.-S., Stenger, B., Yang, M.-H., and Li, H. (2022). Deep image deblurring: A survey. International Journal of Computer Vision, 130(9):2103– 2130.
Zhang, K., Zuo, W., Chen, Y., Meng, D., and Zhang, L. (2017). Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising. IEEE transactions on image processing, 26(7):3142–3155.
Zhang, Z., Chen, Z., Hu, D., Li, M., Xu, Z., Feng, H., Li, Q., and Chen, Y. (2025). Jitter-Aware Restoration With Equivalent Jitter Model for Remote Sensing Push-Broom Image. IEEE Transactions on Geoscience and Remote Sensing.
Zhaoxiang, Z., Iwasaki, A., and Xu, G. (2019). Attitude jitter compensation for remote sensing images using convolutional neural network. IEEE Geoscience and Remote Sensing Letters, 16(9):1358–1362.
Zheng, Y., Zhou, Z., and Huang, H. (2020). A multi-frequency MIMO control method for the 6DOF micro-vibration exciting system. Acta Astronautica, 170:552–569.
Zhu, B., Liu, J. Z., Cauley, S. F., Rosen, B. R., and Rosen, M. S. (2018). Image reconstructionby domain-transform manifold learning. Nature, 555(7697):487–492.
張立雨, 陳良健. (1997). 軌道修正及光線追蹤應用於 SPOT 衛星影像正射化. 航測及遙測學刊, 2(1):43–60.