Super Resolution
Combine multiple aligned scans to produce a super-resolved image with higher spatial resolution. Sub-pixel shifts between inputs are estimated via FFT cross-correlation and used to reconstruct a finer grid. Equivalent to Gwyddion's superresolution.c.
Inputs
| Name |
Type |
Required |
Description |
| field1 |
DATA_FIELD |
Yes |
Reference image |
| field2 |
DATA_FIELD |
No |
Second image |
| field3 |
DATA_FIELD |
No |
Third image |
| field4 |
DATA_FIELD |
No |
Fourth image |
Outputs
| Name |
Type |
Description |
| result |
DATA_FIELD |
Super-resolved image |
Controls
| Name |
Type |
Default |
Description |
| upscale |
INT |
2 |
Upscaling factor (2, 3, or 4) |
Notes
- When only one field is provided the image is upsampled using cubic interpolation (no multi-image enhancement).
- Additional fields are aligned to the reference using FFT-based cross-correlation with parabolic sub-pixel refinement, then averaged on the high-resolution grid.
- Providing more images generally improves the result because each scan samples slightly different sub-pixel positions.
- All input fields should have the same pixel dimensions and physical size for correct alignment.
- The upscale factor controls the output resolution multiplier (2x, 3x, or 4x the input dimensions).
