Products›Image Restoration
Image Restoration
SNUAILAB Denoising Solution
SIRNet-based image restoration reduces the noise and blur that arise in low-dose imaging, improving the discernibility of fine structures and the usability of images needed for battery inspection and medical imaging.
Image Restoration
In precision manufacturing and medical diagnosis,
low-dose imaging matters more than ever
DR (Digital Radiology) is an imaging technique that converts transmitted radiation signals into digital form.
Medical settings and battery inspection processes are divided into Static and Dynamic by imaging method, and each faces its own low-dose noise dilemma.
-
Static DR(one precise shot taken in a fixed position)
When the imaging dose cannot be fully secured or is lowered, noise builds up until fine structures can no longer be distinguished, creating a risk of diagnostic and judgment errors.
Single exposure→Single image acquisition→Single-view capture→Still image
-
Dynamic DR(video technique capturing many frames continuously in real time during procedures or line runs)
Because exposure lasts a long time, imaging must use an ultra-low dose (1K-1.5K) only, and the severe quantum mottle noise makes real-time restoration essential.
Continuous exposure (1K-1.5K fps)→Continuous image acquisition→Continuous-view capture (real time)→High-quality video from continuous frames
Image Restoration
Proprietary SIRNet with SOTA performance
that gives up neither speed nor quality
| Item | Conventional CNN-based models | SNUAILAB SIRNet |
|---|---|---|
| Inference speed | Slow due to deep network structure | Fast inference via optimized parallel computing, real-time ready |
| Computational complexity | Excessive convolution operations | Self-Attention based |
| GPU usage | Excessive GPU memory use | Greatly reduced GPU memory use |
| Restoration quality | Limited restoration of fine structures | Restores even fine structures accurately via global context |
- PSNR28.1 → 33.7 dB
- SSIM0.842 → 0.927
- Inference speed3.2 → 12.8 fps
- GPU memory11.2GB → 4.1GB
01Feature
Low-dose image
denoising
Removes noise caused by low-dose imaging and equipment characteristics to restore the original structure clearly.
02Feature
Still and real-time
image restoration
From single X-ray images to continuous video on production lines and in medical settings, restoration runs in real time to suit the imaging method.
03Feature
Industrial battery
X-ray restoration
Removing noise from battery X-ray images makes key inspection areas such as electrode alignment, internal winding structure, and tab shape clearer to examine.
04Feature
Medical X-ray and CT
image restoration
Reduces noise and afterimages in low-dose X-ray, CT, and real-time intraoperative fluoroscopy, improving the visibility of tissue boundaries and medical device shapes.
Why SNUAILAB?

World-class core technology
Our proprietary SIRNet backbone achieves SOTA-level performance in denoising.

Trainable without ground truth
Restores precisely with a self-supervised approach, even without paired real noise data.

Decisive quantitative advantage
With 178× faster inference and 90% less GPU memory, real-time use in the field is possible.

Proven industry references
Deployed in practice in batteries (Company L, Company S) and medical imaging (Company V).
Experience clarity without blur,
even in high-speed, low-dose conditions
More precise image restoration
starts with SNUAILAB's SIRNet IR
Maximize your business values
by adopting industry-specific AI solutions.