- MATLAB 100%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| examples | ||
| test_image | ||
| adaptive_blind_deblur_test.p | ||
| adjust_psf_center.p | ||
| bilateral_filter.p | ||
| blind_deconv.p | ||
| blind_deconv_main.p | ||
| blind_main_fei.p | ||
| blind_psf_fei.p | ||
| conjgrad.p | ||
| deblur_test08.p | ||
| deblur_tv_pmpr.p | ||
| deblurring_adm_aniso.p | ||
| dfilters.p | ||
| dmaxflat.p | ||
| estimate_psf.p | ||
| find_min_pixels.p | ||
| gradient_1.p | ||
| IMAGE_DEBLUR.m | ||
| IntensityImageFast.p | ||
| L0Deblur_whole.p | ||
| L0Restoration.p | ||
| LICENSE | ||
| MakeONFilter.p | ||
| mctrans.p | ||
| MirrorFilt.p | ||
| modulate2.p | ||
| opt_fft_size.p | ||
| padImage.p | ||
| pcodeLog.txt | ||
| README.md | ||
| ringing_artifacts_removal.p | ||
| SLcheckFilterSizes.p | ||
| SLdshear.p | ||
| SLfiltersToGPU2D.p | ||
| SLgetShearletIdxs2D.p | ||
| SLgetShearlets2D.p | ||
| SLgetShearletSystem2D.p | ||
| SLgetSubsystem2D.p | ||
| SLgetWedgeBandpassAndLowpassFilters2D.p | ||
| SLpadArray.p | ||
| SLprepareFilters2D.p | ||
| SLsheardec2D.p | ||
| SLupsample.p | ||
| test.p | ||
| test2.p | ||
| test_kbregion.p | ||
| test_kernel.p | ||
| threshold_pxpy_v1.p | ||
| wrap_boundary_liu.p | ||
Enhanced Space-Variant Deblurring of Spacecraft Images via Detail-Preserving Techniques
During spacecraft launch, flight, and docking, ground-based and space-based imaging systems play a crucial role in monitoring spacecraft pose and flight status. However, images captured by these systems often suffer from space-variant blurring caused by atmospheric turbulence, defocus, and relative motion. To address this challenge, we propose a novel deblurring method tailored for space variant degraded spacecraft images. Our approach begins by identifying and extracting detail feature regions using multiscale morphological gradients and adaptive thresholds, allowing for a preliminary estimation of the blur kernel. We then construct an initial space-variant blur kernel through linear interpolation and refine it by introducing a space-variant blur kernel error term to mitigate inaccura cies. Additionally, shearlet wavelet regularization is incorporated to enhance local detail restoration. A multivariable minimization alternating iteration strategy is employed to solve the degradation model and achieve image deblurring. Experimental results demonstrate that our method outperforms advanced deblurring techniques in preserving intricate details of space-variant degraded spacecraft images.
Update Log
v1.0.0 (2025-12-08): Initial public release, including inference code and pre-trained kernels.
Installation
Install my-project with npm
git clone https://github.com/bsfsf/Image_deblur
cd Image_deblur
matlab -batch "install_SVID"
Running Tests
To run tests, run the following command
matlab -batch "IMAGE_DEBLUR"
Select the image to be processed in the pop-up file selection dialog. After processing is completed, a dialog will pop up again → Click Yes to generate the output file named *_deblurred.png in the same directory.
Screenshots
| Input | Output |
|---|---|
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
Algorithm & Implementation Sketch
| Stage | Core Function | Key Idea |
|---|---|---|
| Detail-feature extraction | blind_psf_fei → blind_main_fei | multi-scale morphological gradient + adaptive threshold (§3.2) |
| Space-variant kernel estimation | estimate_psf + fangzhenkernel | per-region PSF via conjugate-gradient & connectivity filter (§3.2-3.3) |
| Kernel error removal | filterKernelError (L0 on εKf) | suppress ringing caused by initial kernel inaccuracy (§3.3) |
| Shearlets regularisation | SLgetShearletSystem2D → SLsheardec2D | multi-direction / multi-scale prior, soft-threshold per sub-band (§3.4) |
| Final latent image | deblur_test08 (color) | FFT-based deconvolution + TV-L0 fusion & bilateral post-processing (§3.5) |
Contact
Should you have any question, please contact liang.ye@hotmail.com
Dataset
All experiments were conducted on the Shenzhou Series Spacecraft Image Dataset released by China Manned Space Agency (CMSA).
Source: http://www.cmse.gov.cn/
Rights: Open for academic research with attribution.
Citation:
@misc{shenzhou2025dataset,
title={Shenzhou Series Spacecraft Image Dataset},
howpublished={\url{http://www.cmse.gov.cn/}},
year={2025}
}
Citation
If you use this code or the dataset in your research, please cite our paper:
@article{hong2025enhanced,
title={Enhanced Space-Variant Deblurring of Spacecraft Images via Detail-Preserving Techniques},
author={Hong, Hanyu and Guo, Shuai and Liu, Zhiwen and others},
year={2025}
}





