NGMeet Denoising
Applies NGMeet hyperspectral denoising to reduce noise while preserving useful spatial and spectral information.
Overview
Applies NGMeet hyperspectral denoising to reduce noise while preserving useful spatial and spectral information.
Light: preserves more fine detail and applies less aggressive denoising.
Balanced: general-purpose default.
Strong: applies more aggressive noise reduction.
Sigma represents the assumed noise level. Increasing Sigma generally increases the strength of noise suppression.
Spectral Subspace controls the reduced spectral representation used Iterations controls the number of repeated denoising steps. Additional iterations may improve convergence but increase processing time.
- Hyperspectral data contain substantial spatial-spectral noise.
- Noise reduction is needed before classification or spectral analysis.
- Preservation of spectral structure is important.
NGMeet can be computationally intensive for large hyperspectral cubes.
Processing time depends on image size, number of bands, and settings.
Strong denoising may suppress weak spatial structures or subtle spectral features together with noise.
Workflow
1. Load a hyperspectral dataset.
2. Open Filtering & Enhancement.
3. Choose Light, Balanced, or Strong.
4. Adjust advanced parameters if required.
5. Run the denoising operation.
6. Compare image quality and spectral preservation.
7. Use Reset Filters if the result is not desired.