GUI Application
BM3DORNL includes a standalone GUI application for interactive ring artifact removal from tomography data.
Installation
# Install with the main package
pip install bm3dornl[gui]
# Or install separately
pip install bm3dornl-gui
Launching
bm3dornl-gui
Features
Data Loading
Load HDF5 files with interactive tree browser for dataset selection
Load TIFF files (single images or stacks)
Support for 2D and 3D datasets
Automatic data type detection
Visualization
Interactive slice viewer with scroll-wheel navigation
Real-time histogram display
Adjustable window/level (contrast) controls
Side-by-side comparison of original and processed images
Processing
Real-time parameter adjustment
Support for both
genericandstreakmodesLive preview of denoising results
ROI Selection
Shift+drag to select a region of interest
Histogram updates to show ROI statistics
Useful for evaluating local noise characteristics
Export
Export processed data to TIFF format
Export to HDF5 format
Batch export of full stacks
Keyboard Shortcuts
Shortcut |
Action |
|---|---|
Scroll wheel |
Navigate through slices (3D data) or zoom |
Drag |
Pan image |
Shift + Drag |
Select ROI for histogram |
Workflow
Load Data: Click “Open” and select an HDF5 or TIFF file. For HDF5, use the tree browser to select a dataset.
Adjust View: Use window/level controls or scroll to navigate through slices.
Select Parameters: Adjust denoising parameters in the control panel:
Mode:
streakfor ring artifacts,genericfor random noiseSigma: Start low (0.02-0.05) and increase as needed
Other parameters: Usually defaults work well
Process: Click “Process” to apply denoising. Compare with original using the split view.
Evaluate: Use ROI selection (Shift+drag) to check noise statistics in specific regions.
Export: Save processed data to TIFF or HDF5 format.
Tips
Start with defaults: The default parameters work well for most cases.
Use streak mode: For ring artifacts in sinograms, always use
mode="streak".Check the histogram: The histogram should narrow after denoising, indicating reduced noise.
Use ROI selection: Select flat regions to evaluate noise reduction without signal interference.
Monitor the difference: The removed signal should show primarily vertical streaks (for streak mode), not horizontal structure.