Research map / Image generation / Image restoration
Blur Invariants for Image Recognition
Jan Flusser, Matěj Lébl, Filip Šroubek, Matteo Pedone, Jitka Kostková
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use blur invariants instead of ResNet18 trained on clear images, circular 15x15 blur | Blur invariants · YaleB face recognition (SNR 50 dB, circular 15x15) | recognition rate (%): 35 -> 100 | Improved |
| use blur invariants instead of ResNet18 trained on clear images, circular 10x10 blur | Blur invariants · YaleB face recognition (SNR 50 dB, circular 10x10) | recognition rate (%): 90 -> 100 | Improved |
| use blur invariants instead of Gopalan's distance, circular 125x125, SNR 50 | Blur invariants · YaleB face recognition (SNR 50 dB, circular 125x125) | recognition rate (%): 40 -> 100 | Improved |
| use blur invariants instead of Gopalan's distance, SNR 5, circular 125x125 | Blur invariants · YaleB face recognition (SNR 5 dB, circular 125x125) | recognition rate (%): 5 -> 99,9 | Improved |
| add blur augmentation (100 instances) to CNN training instead of using invariants, motion blur 125x125 | ResNet18 CNN · YaleB face recognition (SNR 5 dB, motion 125x125) | recognition rate (%): 92 -> 99,6 | Improved |
| augmented CNN versus invariants, Gaussian blur 125x125 SNR 5 | ResNet18 CNN · YaleB face recognition (SNR 5 dB, Gaussian 125x125) | recognition rate (%): 91 -> 99,5 | Improved |
| augment CNN training with blurred images versus clean-only training (ablation) | ResNet18 CNN · YaleB face recognition (SNR 50 dB, circular 15x15) | recognition rate (%): 35 -> 100 | Improved |
| use blur invariants instead of plain moments for template matching (ablation) | Blur invariants · Template matching (21 templates, defocused scene) | localization error | Improved |
| increase moment order of invariant | Blur moment invariants · Mean relative error under blur and noise | MRE | Hurt |
Closest papers by meaning
- Normalized Blind Deconvolution (ECCV 2018 · similarity 0.84)
- Spatially-Varying Blur Detection Based on Multiscale Fused and Sorted Transform Coefficients of Gradient Magnitudes (CVPR 2017 · similarity 0.83)
- Blur2Blur: Blur Conversion for Unsupervised Image Deblurring on Unknown Domains (CVPR 2024 · similarity 0.82)
- A Physics-Informed Blur Learning Framework for Imaging Systems (CVPR 2025 · similarity 0.81)
- Real-World Efficient Blind Motion Deblurring via Blur Pixel Discretization (CVPR 2024 · similarity 0.81)
- Selective Disk Bispectrum: A Complete and Rotation Invariant Image Descriptor (NeurIPS 2026 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
