A PyQt6-based GUI application for applying baseline corrections and noise reduction to spectroscopy data.

• Multiple Baseline Correction Algorithms: Access to 50+ baseline correction algorithms from the pybaselines library, with library's original categories:

  • Polynomial (poly, modpoly, imodpoly, etc.)
  • Whittaker (asls, iasls, airpls, etc.)
  • Morphological (mor, imor, rolling_ball, etc.)
  • Spline (mixture_model, pspline variants, etc.)
  • Smoothing (snip, swima, ipsa, etc.)
  • Classification (dietrich, golotvin, fastchrom, etc.)
  • Miscellaneous (beads, custom_bc, etc.)

• Noise Reduction: Savitzky-Golay filter with configurable parameters

• Real-time Visualization: Interactive plot

• Flexible Data Input:

  • Load NumPy binary files (*.npy)
  • Optionally, load xrflab related files (*.atx, *.pck)
  • Generate synthetic spectra for testing

• Dynamic Parameter Adjustment: All algorithm parameters are automatically generated from function signatures with appropriate input widgets

• Python 3.12 or higher
• poetry package manager (pip if poetry was not available)

poetry install

If poetry was not available:

pip install .

If the user has access to xrflab library, installing it would give the user more options for opening spectra files.

poetry add xrflab_url

python GUI.py

1. Load or Generate Data

   • A synthetic spectrum is generated when the app is loaded
   • Click "Open Spectrum" to load existing data files
   • Click "Generate Random Spectrum" to generate random spectrum

2. Configure Baseline Correction

   • Enable/disable baseline correction with the checkbox
   • Select a category from the dropdown
   • Choose a specific algorithm
   • Adjust algorithm parameters as needed

3. Configure Noise Reduction (Optional)

   • Enable noise reduction with the checkbox
   • Adjust Savitzky-Golay filter parameters:
     • Window length (number of points)
     • Polynomial order
     • Derivative order
     • Extension mode

4. View Results

   • The plot updates automatically as you adjust parameters
   • Blue line: Original spectrum
   • Green line: Computed baseline
   • Orange line: Corrected spectrum

If xrflab is available, the below extensions are also supported:

The application is organized into modular components:

• MainWindow: Main application window and UI orchestration
• SpectrumGenerator: Generates synthetic test spectra with Gaussian peaks
• BaselineProcessor: Manages baseline correction algorithms and parameters
• NoiseReducer: Applies Savitzky-Golay filtering
• PlotManager: Handles plot visualization and updates
• FileManager: Manages file I/O operations
• ParameterWidgetFactory: Creates appropriate widgets for algorithm parameters
• ConditionDialog: Dialog for selecting spectrum conditions from multi-condition files

Best for simple, smooth baselines. Uses polynomial fitting with various robustness improvements.

Recommended starting points:

• modpoly: Modified polynomial, good general-purpose algorithm
• imodpoly: Improved modpoly with better peak handling

Penalized least squares approaches. Excellent for complex baselines.

Recommended starting points:

• asls: Asymmetric least squares, very popular
• arpls: Adaptive reweighted penalized least squares

Based on mathematical morphology operations. Good for uneven baselines.

Recommended starting points:

• mor: Basic morphological method
• rolling_ball: Simulates rolling a ball under the spectrum

Use spline interpolation for smooth baseline estimation.

Recommended starting points:

• mixture_model: Statistical mixture model approach
• pspline_asls: P-spline version of ASLS

Apply smoothing operations to estimate baseline.

Recommended starting points:

• snip: Statistics-sensitive non-linear iterative peak-clipping
• noise_median: Median-based smoothing

Use signal classification to identify baseline regions.

Recommended starting points:

• dietrich: Simple and fast classification method
• std_distribution: Standard deviation-based classification

1. Start Simple: Begin with polynomial methods like modpoly before trying more complex algorithms

2. Adjust Lambda/Alpha: For Whittaker methods, the smoothing parameter (usually lam) controls baseline smoothness:

   • Smaller values (1e3 - 1e5): More flexible baseline
   • Larger values (1e6 - 1e8): Smoother baseline

3. Peak Sensitivity: For spectra with sharp peaks, use methods with asymmetric weighting (asls, arpls, etc.)

4. Iterative Refinement: Some algorithms (imodpoly, iasls) iterate to improve results. Adjust iteration parameters if needed.

1. Window Length: Should be odd and larger than polynomial order

   • Smaller windows: Preserve peak shape but less smoothing
   • Larger windows: More smoothing but can distort peaks

2. Polynomial Order:

   • Order 2-3: Good for most spectra
   • Higher orders: Can introduce artifacts

• Parameter changes are debounced (300ms delay) to prevent excessive recalculation
• For large datasets, start with fast algorithms (polynomial, morphological)
• Disable real-time updates if working with very large spectra

Error: "Algorithm 'xxx' not found"

• Ensure pybaselines is properly installed
• Some algorithms may require specific dependencies

Error: "window_length must be less than or equal to the size of x"

• Reduce the window length parameter in noise reduction
• Ensure your spectrum has enough data points

Error: "polyorder must be less than window_length"

• Reduce polynomial order or increase window length

Plot shows error message

• Check parameter values are appropriate for your data
• Try a different algorithm
• Verify input data is valid (no NaN or infinite values)

If you encounter issues:

1. Check parameter values are within reasonable ranges
2. Try the default synthetic spectrum to isolate data vs. algorithm issues
3. Consult the pybaselines documentation for algorithm-specific guidance

This project uses the following open-source libraries:

• PyQt6: GPL v3
• pybaselines: BSD 3-Clause License
• pyqtgraph: MIT License
• NumPy, SciPy: BSD License

Please ensure compliance with respective licenses for distribution.

Contributions are welcome! Areas for improvement:

• Additional baseline correction methods
• Export functionality for corrected spectra
• Batch processing capabilities
• Parameter presets/profiles
• Undo/redo functionality
• Additional visualization options (residuals, difference plots)

• pybaselines: Official Documentation
• Savitzky-Golay filter: SciPy.savgol_filter

Author: [Bahador Rousta Jorshary]
Last Updated: January 2026